fix(health): не показывать Down при выключенном health-check

Маскируем устаревший down, если HC выкл (binding/группа), IP или сервис отключены.
Бейдж Disabled — нейтральный, не красный.

Co-authored-by: Cursor <[email protected]>

apps/api/src/app.ts

@@ -38,6 +38,10 @@ import {
   healthEngineFallbacksFromConfig,
   scheduleHealthCheckJob,
 } from "./services/health-check-scheduler.js";
+import {
+  createWeightedDnsTask,
+  scheduleWeightedDnsJob,
+} from "./services/weighted-dns-scheduler.js";
 import { fireEnsureHealthWorker } from "./services/health/health-worker-deploy.js";
 import { AsyncTask, CronJob } from "toad-scheduler";
 
@@ -133,6 +137,7 @@ export async function buildApp(opts: BuildAppOptions = {}) {
     app.decorate("reloadHealthCheckJob", () => {
       scheduleHealthCheckJob(app, config, healthTask);
     });
+    scheduleWeightedDnsJob(app, createWeightedDnsTask(app));
     if (config.cloudflareApiToken) {
       fireEnsureHealthWorker(
         app.db,

apps/api/src/routes/services.ts

@@ -78,6 +78,18 @@ export async function serviceRoutes(app: FastifyInstance) {
     };
   });
 
+  app.get("/services/:id/failover-log", async (request) => {
+    const { id } = request.params as { id: string };
+    repos.getService(request.server.db, Number(id));
+    return {
+      items: repos.listFailoverLogForService(
+        request.server.db,
+        Number(id),
+        200,
+      ),
+    };
+  });
+
   app.get("/services/:id/certificates", async (request) => {
     const { id } = request.params as { id: string };
     return certificateService.listServiceCertificates(

apps/api/src/services/dns-service.ts

@@ -1,11 +1,17 @@
 import type { Db } from "@cfdm/db";
 import { repos, type DnsListFilter } from "@cfdm/db";
-import type { CreateDnsRecordPayload, DnsRecord, PatchDnsRecordPayload } from "@cfdm/shared";
+import type {
+  CfDnsRecord,
+  CreateDnsRecordPayload,
+  DnsRecord,
+  PatchDnsRecordPayload,
+} from "@cfdm/shared";
 import {
   SYNC_CONFLICT,
   SYNC_ERROR,
   SYNC_PENDING_PUSH,
   SYNC_SYNCED,
+  dnsRecordNamesMatch,
   normalizeDnsRecordName,
 } from "@cfdm/shared";
 import type { CloudflareClient } from "../lib/cf-client.js";
@@ -64,6 +70,86 @@ function toCfPayload(
   };
 }
 
+function isMissingCfDnsRecord(error: unknown): boolean {
+  const message = error instanceof Error ? error.message : String(error);
+  return /record does not exist|81044/i.test(message);
+}
+
+function dnsContentMatches(
+  recordType: string,
+  left: string,
+  right: string,
+): boolean {
+  if (recordType.toUpperCase() === "CNAME") {
+    return (
+      left.trim().replace(/\.+$/, "").toLowerCase() ===
+      right.trim().replace(/\.+$/, "").toLowerCase()
+    );
+  }
+  return left === right;
+}
+
+/** Resolve live Cloudflare record by name + type + content (IP / CNAME target). */
+function findRemoteByIdentity(
+  remote: readonly CfDnsRecord[],
+  zoneName: string,
+  recordType: string,
+  name: string,
+  content: string,
+): CfDnsRecord | undefined {
+  const type = recordType.toUpperCase();
+  return remote.find(
+    (record) =>
+      Boolean(record.id) &&
+      (record.type ?? "").toUpperCase() === type &&
+      dnsRecordNamesMatch(record.name, name, zoneName) &&
+      dnsContentMatches(type, record.content, content),
+  );
+}
+
+function findRemoteByCfId(
+  remote: readonly CfDnsRecord[],
+  cfRecordId: string | null | undefined,
+): CfDnsRecord | undefined {
+  if (!cfRecordId) return undefined;
+  return remote.find((record) => record.id === cfRecordId);
+}
+
+async function markSynced(
+  db: Db,
+  domainId: number,
+  record: DnsRecord,
+  cfRec: {
+    id?: string | null;
+    type?: string;
+    name: string;
+    content: string;
+    ttl: number;
+    proxied?: boolean | null;
+    priority?: number | null;
+  },
+): Promise<DnsRecord> {
+  repos.updateDnsFields(
+    db,
+    record.id,
+    cfRec.type ?? record.record_type,
+    cfRec.name,
+    cfRec.content,
+    cfRec.ttl,
+    cfRec.proxied ?? false,
+    cfRec.priority ?? null,
+    SYNC_SYNCED,
+    cfRec.id ?? null,
+    null,
+  );
+  return repos.getDnsRecord(db, domainId, record.id);
+}
+
+/**
+ * Push local desired state to Cloudflare.
+ * Identity is name + type + content; cf_record_id is only a cache hint
+ * (records may be deleted/recreated outside CFDM).
+ */
 async function pushRecord(
   db: Db,
   cf: CloudflareClient,
@@ -71,6 +157,7 @@ async function pushRecord(
   cfZoneId: string,
   record: DnsRecord,
 ): Promise<DnsRecord> {
+  const domain = repos.getDomain(db, domainId);
   const payload = toCfPayload(
     record.record_type,
     record.name,
@@ -81,25 +168,38 @@ async function pushRecord(
   );
 
   try {
-    const cfRec = record.cf_record_id
-      ? await cf.updateDnsRecord(cfZoneId, record.cf_record_id, payload)
-      : await cf.createDnsRecord(cfZoneId, payload);
-
-    repos.updateDnsFields(
-      db,
-      record.id,
-      cfRec.type ?? record.record_type,
-      cfRec.name,
-      cfRec.content,
-      cfRec.ttl,
-      cfRec.proxied ?? false,
-      cfRec.priority ?? null,
-      SYNC_SYNCED,
-      cfRec.id ?? null,
-      null,
+    const remote = await cf.listDnsRecords(cfZoneId);
+    const byIdentity = findRemoteByIdentity(
+      remote,
+      domain.zone_name,
+      record.record_type,
+      record.name,
+      record.content,
     );
-    return repos.getDnsRecord(db, domainId, record.id);
+    const byCachedId = findRemoteByCfId(remote, record.cf_record_id);
+    const targetId = byIdentity?.id ?? byCachedId?.id ?? null;
+
+    const cfRec = targetId
+      ? await cf.updateDnsRecord(cfZoneId, targetId, payload)
+      : await cf.createDnsRecord(cfZoneId, payload);
+    return markSynced(db, domainId, record, cfRec);
   } catch (e) {
+    // Race: cached id vanished mid-flight — recreate by identity.
+    if (isMissingCfDnsRecord(e)) {
+      try {
+        const created = await cf.createDnsRecord(cfZoneId, payload);
+        return markSynced(db, domainId, record, created);
+      } catch (createErr) {
+        repos.setDnsSyncStatus(
+          db,
+          record.id,
+          SYNC_ERROR,
+          null,
+          createErr instanceof Error ? createErr.message : String(createErr),
+        );
+        throw createErr;
+      }
+    }
     repos.setDnsSyncStatus(
       db,
       record.id,
@@ -188,15 +288,23 @@ export async function patchContent(
 ): Promise<DnsRecord> {
   const domain = repos.getDomain(db, domainId);
   const existing = repos.getDnsRecord(db, domainId, recordId);
-  if (!existing.cf_record_id) {
-    throw AppError.dnsUpdateFailed("у DNS-записи нет идентификатора Cloudflare");
-  }
   try {
-    const cfRec = await cf.patchDnsRecord(
-      domain.cf_zone_id,
-      existing.cf_record_id,
-      payload,
+    const remote = await cf.listDnsRecords(domain.cf_zone_id);
+    const byIdentity = findRemoteByIdentity(
+      remote,
+      domain.zone_name,
+      existing.record_type,
+      existing.name,
+      existing.content,
     );
+    const byCachedId = findRemoteByCfId(remote, existing.cf_record_id);
+    const targetId = byIdentity?.id ?? byCachedId?.id ?? null;
+    if (!targetId) {
+      throw AppError.dnsUpdateFailed(
+        "DNS-запись не найдена в Cloudflare по имени и содержимому",
+      );
+    }
+    const cfRec = await cf.patchDnsRecord(domain.cf_zone_id, targetId, payload);
     repos.updateDnsFields(
       db,
       existing.id,
@@ -207,7 +315,7 @@ export async function patchContent(
       cfRec.proxied ?? existing.proxied,
       cfRec.priority ?? existing.priority,
       SYNC_SYNCED,
-      cfRec.id ?? existing.cf_record_id,
+      cfRec.id ?? targetId,
       null,
     );
     return repos.getDnsRecord(db, domainId, existing.id);
@@ -235,20 +343,41 @@ export async function deleteRecord(
   const record = repos.getDnsRecord(db, domainId, recordId);
   repos.markDnsPendingDelete(db, recordId);
 
-  if (record.cf_record_id) {
+  let targetId: string | null = null;
+  try {
+    const remote = await cf.listDnsRecords(domain.cf_zone_id);
+    const byIdentity = findRemoteByIdentity(
+      remote,
+      domain.zone_name,
+      record.record_type,
+      record.name,
+      record.content,
+    );
+    const byCachedId = findRemoteByCfId(remote, record.cf_record_id);
+    targetId = byIdentity?.id ?? byCachedId?.id ?? null;
+  } catch {
+    // Zone list failed — fall back to cached id only.
+    targetId = record.cf_record_id;
+  }
+
+  if (targetId) {
     try {
-      await cf.deleteDnsRecord(domain.cf_zone_id, record.cf_record_id);
+      await cf.deleteDnsRecord(domain.cf_zone_id, targetId);
     } catch (e) {
-      repos.setDnsSyncStatus(
-        db,
-        recordId,
-        SYNC_ERROR,
-        record.cf_record_id,
-        e instanceof Error ? e.message : String(e),
-      );
-      throw e;
+      // Already gone in Cloudflare (manual delete) — drop local row.
+      if (!isMissingCfDnsRecord(e)) {
+        repos.setDnsSyncStatus(
+          db,
+          recordId,
+          SYNC_ERROR,
+          targetId,
+          e instanceof Error ? e.message : String(e),
+        );
+        throw e;
+      }
     }
   }
+
   repos.deleteDnsRecord(db, recordId);
 }
 
@@ -326,24 +455,30 @@ export async function resolveConflict(
   }
 
   if (req.source === "cloudflare") {
-    if (record.cf_record_id) {
-      const remote = await cf.listDnsRecords(domain.cf_zone_id);
-      const r = remote.find((x) => x.id === record.cf_record_id);
-      if (r) {
-        repos.updateDnsFields(
-          db,
-          recordId,
-          r.type,
-          r.name,
-          r.content,
-          r.ttl,
-          r.proxied ?? false,
-          r.priority ?? null,
-          SYNC_SYNCED,
-          r.id ?? null,
-          null,
-        );
-      }
+    const remote = await cf.listDnsRecords(domain.cf_zone_id);
+    const byIdentity = findRemoteByIdentity(
+      remote,
+      domain.zone_name,
+      record.record_type,
+      record.name,
+      record.content,
+    );
+    const byCachedId = findRemoteByCfId(remote, record.cf_record_id);
+    const r = byIdentity ?? byCachedId;
+    if (r) {
+      repos.updateDnsFields(
+        db,
+        recordId,
+        r.type,
+        r.name,
+        r.content,
+        r.ttl,
+        r.proxied ?? false,
+        r.priority ?? null,
+        SYNC_SYNCED,
+        r.id ?? null,
+        null,
+      );
     }
     return repos.getDnsRecord(db, domainId, recordId);
   }

apps/api/src/services/health-check-service.ts

@@ -290,7 +290,7 @@ export interface RunAllChecksOptions {
     target: HealthCheckTarget,
     prevState: IpHealthState | null,
     nextState: IpHealthState,
-  ) => void;
+  ) => void | Promise<void>;
 }
 
 function sleep(ms: number): Promise<void> {
@@ -326,7 +326,7 @@ function logSourceResult(
   });
 }
 
-function applyAggregatedStatus(
+async function applyAggregatedStatus(
   db: Db,
   target: HealthCheckTarget,
   sources: Array<{ provider: HealthCheckProvider; result: ProbeResult }>,
@@ -386,7 +386,8 @@ function applyAggregatedStatus(
     { colo, provider: statusProvider },
   );
   const matchedNode = repos.findNodeByIp(db, target.ip);
-  if (matchedNode && matchedNode.enabled) {
+  // Binding-scope only: group apply must not clobber node with its own fetch failed.
+  if (matchedNode && matchedNode.enabled && target.scope === "binding") {
     repos.updateNode(db, matchedNode.id, {
       health_status: node,
       consecutive_failures: failures,
@@ -396,7 +397,7 @@ function applyAggregatedStatus(
     });
   }
   if (prevState !== state) {
-    options.onStatusChange?.(target, prevState, state);
+    await options.onStatusChange?.(target, prevState, state);
   }
 }
 
@@ -512,7 +513,7 @@ export async function runAllChecks(
       for (const source of sources) {
         logSourceResult(db, target, source.provider, source.result);
       }
-      applyAggregatedStatus(db, target, sources, options);
+      await applyAggregatedStatus(db, target, sources, options);
     }
   }
 

apps/api/src/services/node-service.ts

@@ -9,7 +9,10 @@ import type {
 import { AppError } from "../errors.js";
 import { isValidIpv4 } from "../lib/validators.js";
 import { getView } from "./service-config-service.js";
-import { selectActiveIpsByMode } from "./routing/index.js";
+import {
+  isSharedPool,
+  resolveDesiredAIps,
+} from "./routing/index.js";
 
 function assertAddress(address: string): void {
   if (!isValidIpv4(address)) {
@@ -89,8 +92,12 @@ export async function getOverview(
       : null;
 
   const active = new Set<string>();
+  const serviceIps = service.ips.filter(
+    (ip) => service.ip_enabled[ip] !== false,
+  );
   for (const binding of bindings) {
     const metas = repos.listBindingIpsWithMeta(db, binding.id);
+    const targetIps = metas.map((entry) => entry.ip);
     const rows = metas.map((entry) => {
       const status = repos.getIpHealthStatusRow(db, "binding", binding.id, entry.ip);
       return {
@@ -100,12 +107,15 @@ export async function getOverview(
         health: status ? status.status : ("unknown" as const),
       };
     });
-    for (const ip of selectActiveIpsByMode(
+    for (const ip of resolveDesiredAIps(
       {
         lb_mode: binding.lb_mode,
         health_check_enabled: binding.health_check_enabled,
       },
       rows,
+      targetIps,
+      Date.now(),
+      serviceIps,
     )) {
       active.add(ip);
     }
@@ -134,10 +144,13 @@ export function opsSummary(db: Db) {
     if (binding.lb_mode !== "failover" || !binding.health_check_enabled) {
       return false;
     }
-    return binding.target_ips.some((ip) => {
-      const row = repos.getIpHealthStatusRow(db, "binding", binding.id, ip);
-      return row?.status === "down";
-    });
+    return (
+      isSharedPool(binding.target_ips) &&
+      binding.target_ips.some((ip) => {
+        const row = repos.getIpHealthStatusRow(db, "binding", binding.id, ip);
+        return row?.status === "down";
+      })
+    );
   }).length;
   return {
     domains: domains.length,

apps/api/src/services/routing/failover.ts

@@ -1,16 +1,16 @@
 import type { LbIpRow } from "./types.js";
-import { isHealthy } from "./health.js";
+import { isPoolMember } from "./health.js";
 
 export function failoverDesired(rows: LbIpRow[]): string[] {
   if (rows.length === 0) return [];
-  const healthy = rows.filter((r) => isHealthy(r.health));
-  const pool = healthy.length > 0 ? healthy : rows;
+  const live = rows.filter((r) => isPoolMember(r.health));
+  const pool = live.length > 0 ? live : rows;
   const sorted = [...pool].sort(
     (a, b) => a.priority - b.priority || a.weight - b.weight,
   );
   const minPriority = sorted[0]!.priority;
   const primaries = sorted.filter((r) => r.priority === minPriority);
-  if (healthy.length > 0) {
+  if (live.length > 0) {
     return primaries.map((r) => r.ip);
   }
   return [sorted[0]!.ip];

apps/api/src/services/routing/health.ts

@@ -3,3 +3,12 @@ import type { IpHealthState, NodeHealthState } from "@cfdm/shared";
 export function isHealthy(state: IpHealthState | NodeHealthState | string): boolean {
   return state === "up" || state === "healthy";
 }
+
+export function isDown(state: IpHealthState | NodeHealthState | string): boolean {
+  return state === "down" || state === "unhealthy";
+}
+
+/** A-pool membership: only Down is drained. Recovering (unknown/checking) and Slow return immediately. */
+export function isPoolMember(state: IpHealthState | NodeHealthState | string): boolean {
+  return !isDown(state);
+}

apps/api/src/services/routing/index.ts

@@ -1,26 +1,55 @@
 import type { LbMode } from "@cfdm/shared";
 import { failoverDesired } from "./failover.js";
+import { canApplyLb, isSharedPool } from "./pool.js";
 import { roundRobinDesired } from "./round-robin.js";
 import type { LbIpRow, LbTargetConfig } from "./types.js";
+import { weightedDesired } from "./weighted.js";
 
 export type { LbIpRow, LbTargetConfig } from "./types.js";
-export { isHealthy } from "./health.js";
+export { isDown, isHealthy, isPoolMember } from "./health.js";
 export { withBindingLock } from "./binding-lock.js";
+export {
+  canApplyLb,
+  isSharedPool,
+  shouldRecordFailoverDnsDiff,
+  uniqueIpCount,
+} from "./pool.js";
+export { WEIGHTED_DNS_TTL, WEIGHTED_SLOT_MS, weightedDesired } from "./weighted.js";
 
 export function selectActiveIpsByMode(
   config: LbTargetConfig,
   rows: LbIpRow[],
+  nowMs = Date.now(),
 ): string[] {
   if (rows.length === 0) return [];
   if (config.lb_mode === "failover") {
     return failoverDesired(rows);
   }
-  // weighted = round_robin on DNS (one A per IP)
+  if (config.lb_mode === "weighted") {
+    return weightedDesired(rows, nowMs);
+  }
   return roundRobinDesired(rows);
 }
 
+export function resolveDesiredAIps(
+  config: LbTargetConfig,
+  rows: LbIpRow[],
+  fallbackIps: readonly string[],
+  nowMs = Date.now(),
+  serviceIps: readonly string[] = fallbackIps,
+): string[] {
+  const fallback = [...fallbackIps];
+  if (!canApplyLb(serviceIps, fallback)) return fallback;
+  if (!isSharedPool(rows.map((row) => row.ip))) return fallback;
+  if (config.lb_mode === "weighted" || config.health_check_enabled) {
+    const activeIps = selectActiveIpsByMode(config, rows, nowMs);
+    if (activeIps.length > 0) return activeIps;
+  }
+  return fallback;
+}
+
 export function strategyLabel(mode: LbMode): string {
   if (mode === "failover") return "Failover";
-  if (mode === "weighted") return "Round Robin (weighted alias)";
+  if (mode === "weighted") return "Weighted";
   return "Round Robin";
 }

apps/api/src/services/routing/pool.ts

@@ -0,0 +1,30 @@
+import type { LbMode } from "@cfdm/shared";
+
+export function uniqueIpCount(ips: readonly string[]): number {
+  return new Set(ips.filter(Boolean)).size;
+}
+
+/** Shared pool — two or more unique IPs. One IP (even duplicated) is not a pool. */
+export function isSharedPool(ips: readonly string[]): boolean {
+  return uniqueIpCount(ips) >= 2;
+}
+
+/** LB / drain only when the service itself has a pool AND this FQDN is shared. */
+export function canApplyLb(
+  serviceIps: readonly string[],
+  bindingIps: readonly string[],
+): boolean {
+  return isSharedPool(serviceIps) && isSharedPool(bindingIps);
+}
+
+export function shouldRecordFailoverDnsDiff(input: {
+  configuredIps: readonly string[];
+  lbMode: LbMode;
+  added: readonly string[];
+  removed: readonly string[];
+  downIps: ReadonlySet<string>;
+}): boolean {
+  if (!isSharedPool(input.configuredIps)) return false;
+  if (input.lbMode !== "weighted") return true;
+  return [...input.added, ...input.removed].some((ip) => input.downIps.has(ip));
+}

apps/api/src/services/routing/round-robin.ts

@@ -1,8 +1,8 @@
 import type { LbIpRow } from "./types.js";
-import { isHealthy } from "./health.js";
+import { isPoolMember } from "./health.js";
 
 export function roundRobinDesired(rows: LbIpRow[]): string[] {
-  const healthy = rows.filter((r) => isHealthy(r.health));
-  const pool = healthy.length > 0 ? healthy : rows;
+  const live = rows.filter((r) => isPoolMember(r.health));
+  const pool = live.length > 0 ? live : rows;
   return pool.map((r) => r.ip);
 }

apps/api/src/services/routing/weighted.ts

@@ -0,0 +1,24 @@
+import type { LbIpRow } from "./types.js";
+import { isPoolMember } from "./health.js";
+
+/** Slot length for time-sliced weighted DNS (one A at a time). */
+export const WEIGHTED_SLOT_MS = 60_000;
+
+/** Cloudflare DNS-only minimum TTL; Auto (1) is ~300s and would smear ratios. */
+export const WEIGHTED_DNS_TTL = 60;
+
+export function weightedDesired(rows: LbIpRow[], nowMs = Date.now()): string[] {
+  if (rows.length === 0) return [];
+  const live = rows.filter((r) => isPoolMember(r.health));
+  const pool = live.length > 0 ? live : rows;
+  if (pool.length === 1) return [pool[0]!.ip];
+
+  const sorted = [...pool].sort((a, b) => a.ip.localeCompare(b.ip));
+  const cycle: string[] = [];
+  for (const row of sorted) {
+    const weight = Math.max(1, Math.round(row.weight));
+    for (let i = 0; i < weight; i++) cycle.push(row.ip);
+  }
+  const slot = Math.floor(nowMs / WEIGHTED_SLOT_MS) % cycle.length;
+  return [cycle[slot]!];
+}

apps/api/src/services/service-config-service.ts

@@ -29,14 +29,91 @@ import * as domainService from "./domain-service.js";
 import { syncServiceToVpsTracker } from "./vps-tracker-sync.js";
 import { fireEnsureHealthWorker, DEFAULT_HEALTH_FALLBACKS } from "./health/health-worker-deploy.js";
 import {
+  canApplyLb,
+  isPoolMember,
+  isSharedPool,
+  resolveDesiredAIps,
   selectActiveIpsByMode,
+  shouldRecordFailoverDnsDiff,
   withBindingLock,
+  WEIGHTED_DNS_TTL,
   type LbIpRow,
   type LbTargetConfig,
 } from "./routing/index.js";
 
 export type { LbIpRow, LbTargetConfig };
-export { selectActiveIpsByMode };
+export {
+  canApplyLb,
+  resolveDesiredAIps,
+  selectActiveIpsByMode,
+  shouldRecordFailoverDnsDiff,
+};
+
+const AUTO_DNS_TTL = 1;
+
+function ttlForBinding(mode: LbMode, ips: readonly string[]): number {
+  return mode === "weighted" && isSharedPool(ips) ? WEIGHTED_DNS_TTL : AUTO_DNS_TTL;
+}
+
+function enabledServiceIps(db: Db, serviceId: number): string[] {
+  return repos
+    .listServiceIpRows(db, serviceId)
+    .filter((row) => row.enabled)
+    .map((row) => row.ip);
+}
+
+export function failoverARecordDiff(
+  existingA: readonly string[],
+  desiredIps: readonly string[],
+): { added: string[]; removed: string[] } {
+  const before = new Set(existingA);
+  const after = new Set(desiredIps);
+  return {
+    added: desiredIps.filter((ip) => !before.has(ip)),
+    removed: existingA.filter((ip) => !after.has(ip)),
+  };
+}
+
+function recordFailoverDnsDiff(
+  db: Db,
+  bindingId: number,
+  hostname: string,
+  zoneName: string,
+  existingRecords: DnsRecord[],
+  desiredIps: string[],
+): void {
+  const existingA = existingRecords
+    .filter((record) => record.record_type.toUpperCase() === "A")
+    .map((record) => record.content);
+  const { added, removed } = failoverARecordDiff(existingA, desiredIps);
+  if (added.length === 0 && removed.length === 0) return;
+  const binding = repos.getBinding(db, bindingId);
+  const configuredIps = repos.listBindingIps(db, bindingId);
+  const { config, rows } = getBindingLbState(db, bindingId);
+  const downIps = new Set(
+    rows.filter((row) => row.health === "down").map((row) => row.ip),
+  );
+  if (
+    !shouldRecordFailoverDnsDiff({
+      configuredIps,
+      lbMode: config.lb_mode,
+      added,
+      removed,
+      downIps,
+    })
+  ) {
+    return;
+  }
+  repos.insertFailoverLog(db, {
+    serviceId: binding.service_id,
+    bindingId,
+    fqdn: fqdnToDisplay(hostname, zoneName),
+    entries: [
+      ...added.map((ip) => ({ ip, action: "added" as const })),
+      ...removed.map((ip) => ({ ip, action: "removed" as const })),
+    ],
+  });
+}
 
 export interface ServiceDomainInput {
   fqdn: string;
@@ -210,7 +287,7 @@ function getGroupLbState(
         } else {
           existing.weight += weight;
           existing.priority = Math.min(existing.priority, priority);
-          if (isHealthy(existing.health) && status && !isHealthy(status.status as IpHealthState)) {
+          if (isPoolMember(existing.health) && status && !isPoolMember(status.status as IpHealthState)) {
             existing.health = status.status as IpHealthState;
           }
         }
@@ -226,16 +303,31 @@ function getGroupLbState(
   };
 }
 
-function computeActiveIps(
+function desiredAIps(
   db: Db,
   scope: HealthCheckScope,
   refId: number,
+  fallbackIps: string[],
 ): string[] {
   const state =
     scope === "binding"
       ? getBindingLbState(db, refId)
       : getGroupLbState(db, refId);
-  return selectActiveIpsByMode(state.config, state.rows);
+  // Configured binding/group IPs stay intact; DNS publishes only enabled ones.
+  const enabledIps =
+    scope === "binding"
+      ? enabledServiceIps(db, repos.getBinding(db, refId).service_id)
+      : fallbackIps;
+  const enabledSet = new Set(enabledIps);
+  const activeFallback = fallbackIps.filter((ip) => enabledSet.has(ip));
+  const activeRows = state.rows.filter((row) => enabledSet.has(row.ip));
+  return resolveDesiredAIps(
+    state.config,
+    activeRows,
+    activeFallback,
+    Date.now(),
+    enabledIps,
+  );
 }
 
 async function collectKnownZones(
@@ -253,7 +345,38 @@ async function collectKnownZones(
   return zones;
 }
 
+/** Restore common A-bindings whose IPs were shrunk by legacy IP toggles. */
+function repairPoolSubsetBindings(db: Db, serviceId: number): void {
+  const pool = repos.listServiceIps(db, serviceId);
+  if (pool.length < 2) return;
+  const poolSet = new Set(pool);
+
+  for (const binding of repos.listBindingsByService(db, serviceId)) {
+    if (binding.cname_target?.trim()) continue;
+    const current = repos.listBindingIpsWithMeta(db, binding.id);
+    if (current.length <= 1) continue;
+    if (!current.every((entry) => poolSet.has(entry.ip))) continue;
+
+    const currentSet = new Set(current.map((entry) => entry.ip));
+    if (currentSet.size === pool.length && pool.every((ip) => currentSet.has(ip))) {
+      continue;
+    }
+
+    const byIp = new Map(current.map((entry) => [entry.ip, entry]));
+    repos.replaceBindingIpsWithMeta(
+      db,
+      binding.id,
+      pool.map((ip) => ({
+        ip,
+        weight: byIp.get(ip)?.weight ?? 1,
+        priority: byIp.get(ip)?.priority ?? 1,
+      })),
+    );
+  }
+}
+
 async function buildView(db: Db, serviceId: number): Promise<ServiceView> {
+  repairPoolSubsetBindings(db, serviceId);
   const service = repos.getService(db, serviceId);
   const ipRows = repos.listServiceIpRows(db, serviceId);
   const ips = ipRows.map((row) => row.ip);
@@ -284,6 +407,11 @@ async function buildView(db: Db, serviceId: number): Promise<ServiceView> {
       if (target_ip_priorities[ip] === undefined) target_ip_priorities[ip] = 1;
     }
 
+    const { config, rows } = getBindingLbState(db, binding.id);
+    const bindingActiveIps = targetCname
+      ? []
+      : resolveDesiredAIps(config, rows, targetIps, Date.now(), ips);
+
     return {
       binding_id: binding.id,
       domain_id: binding.domain_id,
@@ -311,13 +439,13 @@ async function buildView(db: Db, serviceId: number): Promise<ServiceView> {
       health_check_aggregate: binding.health_check_aggregate ?? "majority",
       cert_monitoring: binding.cert_monitoring ?? "auto",
       sync_status: aggregateSyncStatus(statuses),
+      active_ips: bindingActiveIps,
     };
   });
 
   const activeIps = new Set<string>();
-  for (const binding of bindings) {
-    const { config, rows } = getBindingLbState(db, binding.id);
-    for (const ip of selectActiveIpsByMode(config, rows)) {
+  for (const domain of domainViews) {
+    for (const ip of domain.active_ips) {
       activeIps.add(ip);
     }
   }
@@ -403,6 +531,49 @@ function fallbackCnameHealth(
   );
 }
 
+function overlayLiveHealth(
+  stored: IpHealthState | undefined,
+  live: IpHealthState | undefined,
+): IpHealthState {
+  if (live && live !== "unknown") return live;
+  return stored ?? live ?? "unknown";
+}
+
+function bestAliveDisplayStatus(statuses: readonly string[]): IpHealthState {
+  if (statuses.some((status) => status === "up")) return "up";
+  if (statuses.some((status) => status === "degraded")) return "degraded";
+  if (statuses.some((status) => status === "down")) return "down";
+  return "unknown";
+}
+
+/** Health badge applies only when the service, IP and HC (binding or group) are active. */
+function isServiceHealthCheckActive(db: Db, view: ServiceView): boolean {
+  if ((view.domains ?? []).some((domain) => domain.health_check_enabled)) {
+    return true;
+  }
+  if (!view.service_group_id) return false;
+  const group = repos.getServiceGroup(db, view.service_group_id);
+  return Boolean(group.health_check_enabled);
+}
+
+function isIpHealthMonitored(db: Db, view: ServiceView, ip: string): boolean {
+  if (!view.enabled) return false;
+  if (view.ip_enabled[ip] === false) return false;
+  return isServiceHealthCheckActive(db, view);
+}
+
+function inactiveIpHealthRow(ip: string): ServiceHealthRow {
+  return {
+    ip,
+    status: "unknown",
+    latency_ms: null,
+    last_checked_at: null,
+    last_error: null,
+    provider: "local",
+    colo: null,
+  };
+}
+
 function attachServiceHealth(
   db: Db,
   views: ServiceView[],
@@ -410,10 +581,13 @@ function attachServiceHealth(
   const ids = views.map((v) => v.id);
   const healthByService = repos.aggregateIpHealthByServiceIds(db, ids);
   const ipHealthByService = repos.listIpHealthByServiceIds(db, ids);
+  const liveByService = repos.listLatestLiveHealthByServiceIds(db, ids);
   return views.map((view) => {
     const health = healthByService.get(view.id);
     const rows = ipHealthByService.get(view.id) ?? [];
+    const liveRows = liveByService.get(view.id) ?? [];
     const byIp = new Map(rows.map((row) => [row.ip, row]));
+    const liveByIp = new Map(liveRows.map((row) => [row.ip, row]));
     const cnameFallback = fallbackCnameHealth(rows, view);
     const aRecordIps = new Set(
       (view.domains ?? []).flatMap((domain) =>
@@ -421,21 +595,46 @@ function attachServiceHealth(
       ),
     );
     const ip_health = (view.ips ?? []).map((ip) => {
+      if (!isIpHealthMonitored(db, view, ip)) {
+        return inactiveIpHealthRow(ip);
+      }
       const row = byIp.get(ip) ?? (aRecordIps.has(ip) ? undefined : cnameFallback);
+      const live = liveByIp.get(ip);
+      const status = overlayLiveHealth(row?.status, live?.status);
+      const extras = live && live.status !== "unknown" ? live : row;
       return {
         ip,
-        status: row?.status ?? ("unknown" as const),
-        latency_ms: row?.latency_ms ?? null,
-        last_checked_at: row?.last_checked_at ?? null,
-        last_error: row?.last_error ?? null,
-        provider: row?.provider ?? "local",
-        colo: row?.colo ?? null,
+        status,
+        latency_ms: extras?.latency_ms ?? null,
+        last_checked_at: extras?.last_checked_at ?? null,
+        last_error:
+          live && live.status !== "unknown"
+            ? live.last_error
+            : (row?.last_error ?? null),
+        provider: extras?.provider ?? "local",
+        colo: extras?.colo ?? null,
       };
     });
+    const monitoredStatuses = ip_health
+      .filter((row) => isIpHealthMonitored(db, view, row.ip))
+      .map((row) => row.status);
+    const displayStatus =
+      monitoredStatuses.length > 0
+        ? bestAliveDisplayStatus(monitoredStatuses)
+        : ("unknown" as const);
+    const latencyRow =
+      ip_health.find((row) => row.status === displayStatus && row.latency_ms != null) ??
+      ip_health.find((row) => row.latency_ms != null);
     return {
       ...view,
-      health_status: health?.health_status ?? "unknown",
-      health_latency_ms: health?.health_latency_ms ?? null,
+      health_status:
+        monitoredStatuses.length > 0
+          ? overlayLiveHealth(health?.health_status, displayStatus)
+          : "unknown",
+      health_latency_ms:
+        monitoredStatuses.length > 0 && displayStatus !== "unknown"
+          ? (latencyRow?.latency_ms ?? null)
+          : null,
       ip_health,
     };
   });
@@ -538,26 +737,11 @@ async function syncBindingDns(
   desiredIps: string[],
   cnameTarget: string | null,
 ): Promise<void> {
-  const domain = repos.getDomain(db, domainId);
-  const zoneName = domain.zone_name;
-  let effectiveCname = cnameTarget?.trim() || null;
-
-  if (!effectiveCname) {
-    const existingCname = await findOrImportDnsRecord(
-      db,
-      cf,
-      domainId,
-      zoneName,
-      hostname,
-      "CNAME",
-    );
-    if (existingCname) {
-      effectiveCname = existingCname.content;
-      repos.setBindingCnameTarget(db, bindingId, effectiveCname);
-      repos.replaceBindingIps(db, bindingId, []);
-    }
-  }
+  const effectiveCname = cnameTarget?.trim() || null;
 
+  // Desired config wins. Do NOT auto-adopt leftover CNAME from local/CF when the
+  // binding is A-mode — that wiped IPs and blocked extra FQDN publishes.
+  // Docs: https://developers.cloudflare.com/dns/manage-dns-records/troubleshooting/records-with-same-name/
   if (effectiveCname) {
     await syncBindingCnameDns(
       db,
@@ -570,7 +754,83 @@ async function syncBindingDns(
     return;
   }
 
-  await syncBindingADns(db, cf, bindingId, domainId, hostname, desiredIps);
+  const binding = repos.getBinding(db, bindingId);
+  const configuredIps = repos.listBindingIps(db, bindingId);
+  await syncBindingADns(
+    db,
+    cf,
+    bindingId,
+    domainId,
+    hostname,
+    desiredIps,
+    ttlForBinding(binding.lb_mode, configuredIps),
+  );
+}
+
+/** Delete every local+CF record for hostname that must not remain in A mode. */
+async function reconcileHostnameForA(
+  db: Db,
+  cf: CloudflareClient,
+  domainId: number,
+  zoneName: string,
+  hostname: string,
+  desiredIps: readonly string[],
+): Promise<void> {
+  const desired = new Set(desiredIps);
+
+  // CNAME on the same name blocks creating A records in Cloudflare.
+  for (;;) {
+    const cname = await findOrImportDnsRecord(
+      db,
+      cf,
+      domainId,
+      zoneName,
+      hostname,
+      "CNAME",
+    );
+    if (!cname) break;
+    await dnsService.deleteRecord(db, cf, domainId, cname.id);
+  }
+
+  const domain = repos.getDomain(db, domainId);
+  const staleLocal = repos
+    .listDnsByDomain(db, domainId)
+    .filter(
+      (record) =>
+        record.record_type.toUpperCase() === "A" &&
+        dnsRecordNamesMatch(record.name, hostname, zoneName) &&
+        !desired.has(record.content),
+    );
+  for (const record of staleLocal) {
+    await dnsService.deleteRecord(db, cf, domainId, record.id);
+  }
+
+  const remote = await cf.listDnsRecords(domain.cf_zone_id);
+  for (const cfRec of remote) {
+    if ((cfRec.type ?? "").toUpperCase() !== "A" || !cfRec.id) continue;
+    if (!dnsRecordNamesMatch(cfRec.name, hostname, zoneName)) continue;
+    if (desired.has(cfRec.content)) continue;
+
+    const existing = repos.findDnsByCfId(db, domainId, cfRec.id);
+    if (existing) {
+      await dnsService.deleteRecord(db, cf, domainId, existing.id);
+      continue;
+    }
+    const imported = repos.insertDnsRecord(
+      db,
+      domainId,
+      cfRec.type,
+      cfRec.name,
+      cfRec.content,
+      cfRec.ttl,
+      cfRec.proxied ?? false,
+      cfRec.priority ?? null,
+      SYNC_SYNCED,
+      "cloudflare",
+      cfRec.id,
+    );
+    await dnsService.deleteRecord(db, cf, domainId, imported.id);
+  }
 }
 
 async function syncBindingCnameDns(
@@ -584,6 +844,21 @@ async function syncBindingCnameDns(
   const domain = repos.getDomain(db, domainId);
   const zoneName = domain.zone_name;
   const normalized = normalizeCnameTarget(cnameTarget, zoneName);
+
+  // A/AAAA on the same name blocks CNAME create in Cloudflare.
+  for (;;) {
+    const conflictingA = await findOrImportDnsRecord(
+      db,
+      cf,
+      domainId,
+      zoneName,
+      hostname,
+      "A",
+    );
+    if (!conflictingA) break;
+    await dnsService.deleteRecord(db, cf, domainId, conflictingA.id);
+  }
+
   const existingRecords = repos.listRecordsForBinding(db, bindingId);
 
   for (const record of existingRecords) {
@@ -657,9 +932,14 @@ async function syncBindingADns(
   domainId: number,
   hostname: string,
   desiredIps: string[],
+  ttl: number,
 ): Promise<void> {
   const domain = repos.getDomain(db, domainId);
   const zoneName = domain.zone_name;
+
+  // Align zone/local leftovers with desired A set (CNAME conflicts, stale A IPs).
+  await reconcileHostnameForA(db, cf, domainId, zoneName, hostname, desiredIps);
+
   const existingRecords = repos.listRecordsForBinding(db, bindingId);
 
   for (const record of existingRecords) {
@@ -676,6 +956,14 @@ async function syncBindingADns(
 
   if (desiredIps.length === 0) {
     repos.setBindingDnsRecordId(db, bindingId, null);
+    recordFailoverDnsDiff(
+      db,
+      bindingId,
+      hostname,
+      zoneName,
+      existingRecords,
+      desiredIps,
+    );
     return;
   }
 
@@ -684,13 +972,15 @@ async function syncBindingADns(
 
   for (const ip of desiredIps) {
     const existing = refreshed.find((r) => r.content === ip);
+    const recordName = dnsNameForBinding(hostname, zoneName);
     let recordId: number;
     if (existing) {
-      if (!dnsRecordNamesMatch(existing.name, hostname, zoneName)) {
+      if (!dnsRecordNamesMatch(existing.name, hostname, zoneName) || existing.ttl !== ttl) {
         await dnsService.update(db, cf, domainId, existing.id, {
           record_type: "A",
-          name: dnsNameForBinding(hostname, zoneName),
+          name: recordName,
           content: ip,
+          ttl,
           proxied: false,
         });
       }
@@ -708,12 +998,24 @@ async function syncBindingADns(
       if (adopted) {
         repos.linkBindingRecord(db, bindingId, adopted.id);
         recordId = adopted.id;
+        if (
+          !dnsRecordNamesMatch(adopted.name, hostname, zoneName) ||
+          adopted.ttl !== ttl
+        ) {
+          await dnsService.update(db, cf, domainId, adopted.id, {
+            record_type: "A",
+            name: recordName,
+            content: ip,
+            ttl,
+            proxied: false,
+          });
+        }
       } else {
         const record = await dnsService.create(db, cf, domainId, {
           record_type: "A",
-          name: dnsNameForBinding(hostname, zoneName),
+          name: recordName,
           content: ip,
-          ttl: 1,
+          ttl,
           proxied: false,
         });
         repos.linkBindingRecord(db, bindingId, record.id);
@@ -724,6 +1026,14 @@ async function syncBindingADns(
   }
 
   repos.setBindingDnsRecordId(db, bindingId, primaryId);
+  recordFailoverDnsDiff(
+    db,
+    bindingId,
+    hostname,
+    zoneName,
+    existingRecords,
+    desiredIps,
+  );
 }
 
 async function cleanupBindingDns(
@@ -942,12 +1252,7 @@ async function syncServiceBindingsToDns(
     }
     validateTargetIpsInPool(targetIps, ips);
 
-    if (binding.health_check_enabled) {
-      const activeIps = computeActiveIps(db, "binding", binding.id);
-      if (activeIps.length > 0) {
-        targetIps = activeIps;
-      }
-    }
+    const desiredIps = desiredAIps(db, "binding", binding.id, targetIps);
 
     await syncBindingDns(
       db,
@@ -955,7 +1260,7 @@ async function syncServiceBindingsToDns(
       binding.id,
       binding.domain_id,
       binding.hostname,
-      targetIps,
+      desiredIps,
       null,
     );
   }
@@ -969,10 +1274,12 @@ async function collectGroupDnsIps(
   const ips: string[] = [];
   for (const service of services) {
     if (!service.enabled) continue;
+    const enabled = new Set(enabledServiceIps(db, service.id));
     const bindings = repos.listBindingsByService(db, service.id);
     for (const binding of bindings) {
       for (const ip of repos.listBindingIps(db, binding.id)) {
-        if (!ips.includes(ip)) ips.push(ip);
+        if (!enabled.has(ip) || ips.includes(ip)) continue;
+        ips.push(ip);
       }
     }
   }
@@ -987,6 +1294,7 @@ async function syncGroupDomainDnsRecords(
   domainId: number,
   hostname: string,
   desiredIps: string[],
+  ttl: number = AUTO_DNS_TTL,
 ): Promise<void> {
   const domain = repos.getDomain(db, domainId);
   const zoneName = domain.zone_name;
@@ -1002,14 +1310,16 @@ async function syncGroupDomainDnsRecords(
   if (desiredIps.length === 0) return;
 
   const refreshed = repos.listGroupDnsRecords(db, groupId);
+  const recordName = dnsNameForBinding(hostname, zoneName);
   for (const ip of desiredIps) {
     const existing = refreshed.find((r) => r.content === ip);
     if (existing) {
-      if (!dnsRecordNamesMatch(existing.name, hostname, zoneName)) {
+      if (!dnsRecordNamesMatch(existing.name, hostname, zoneName) || existing.ttl !== ttl) {
         await dnsService.update(db, cf, domainId, existing.id, {
           record_type: "A",
-          name: dnsNameForBinding(hostname, zoneName),
+          name: recordName,
           content: ip,
+          ttl,
           proxied: false,
         });
       }
@@ -1025,13 +1335,25 @@ async function syncGroupDomainDnsRecords(
     );
     if (adopted) {
       repos.linkGroupDnsRecord(db, groupId, adopted.id);
+      if (
+        !dnsRecordNamesMatch(adopted.name, hostname, zoneName) ||
+        adopted.ttl !== ttl
+      ) {
+        await dnsService.update(db, cf, domainId, adopted.id, {
+          record_type: "A",
+          name: recordName,
+          content: ip,
+          ttl,
+          proxied: false,
+        });
+      }
       continue;
     }
     const record = await dnsService.create(db, cf, domainId, {
       record_type: "A",
-      name: dnsNameForBinding(hostname, zoneName),
+      name: recordName,
       content: ip,
-      ttl: 1,
+      ttl,
       proxied: false,
     });
     repos.linkGroupDnsRecord(db, groupId, record.id);
@@ -1082,9 +1404,8 @@ async function syncGroupDomainDns(
   const knownZones = await collectKnownZones(db, cf);
   const { zoneName, hostname } = parseFqdn(domainValue, knownZones);
   const domainId = await resolveDomainId(db, cf, zoneName);
-  const desiredIps = group.health_check_enabled
-    ? computeActiveIps(db, "group", groupId)
-    : await collectGroupDnsIps(db, groupId);
+  const fallbackIps = await collectGroupDnsIps(db, groupId);
+  const desiredIps = desiredAIps(db, "group", groupId, fallbackIps);
   await syncGroupDomainDnsRecords(
     db,
     cf,
@@ -1092,6 +1413,7 @@ async function syncGroupDomainDns(
     domainId,
     hostname,
     desiredIps,
+    ttlForBinding(group.lb_mode, fallbackIps),
   );
 }
 
@@ -1272,14 +1594,7 @@ export async function updateConfig(
       }
 
       if (pushDns) {
-        let effectiveIps = targetIps;
-        const refreshedBinding = repos.getBinding(db, binding.id);
-        if (refreshedBinding.health_check_enabled) {
-          const activeIps = computeActiveIps(db, "binding", binding.id);
-          if (activeIps.length > 0) {
-            effectiveIps = activeIps;
-          }
-        }
+        const effectiveIps = desiredAIps(db, "binding", binding.id, targetIps);
         await syncBindingDns(
           db,
           cf,
@@ -1485,26 +1800,8 @@ export async function toggleServiceIp(
     repos.updateNode(db, node.id, { enabled });
   }
 
-  const bindings = repos.listBindingsByService(db, serviceId);
-  for (const binding of bindings) {
-    if (binding.cname_target?.trim()) continue;
-    const current = repos.listBindingIpsWithMeta(db, binding.id);
-    const hasIp = current.some((entry) => entry.ip === ip);
-    if (enabled && !hasIp) {
-      repos.replaceBindingIpsWithMeta(db, binding.id, [
-        ...current,
-        { ip, weight: 1, priority: 1 },
-      ]);
-      continue;
-    }
-    if (!enabled && hasIp) {
-      repos.replaceBindingIpsWithMeta(
-        db,
-        binding.id,
-        current.filter((entry) => entry.ip !== ip),
-      );
-    }
-  }
+  // Keep binding IP membership stable (common FQDN = full pool). DNS sync
+  // filters by enabledServiceIps via desiredAIps — do not reshuffle bindings.
 
   const service = repos.getService(db, serviceId);
   if (shouldPushDns(db, service)) {
@@ -1585,16 +1882,18 @@ export async function reconcileDnsForTarget(
   if (scope === "binding") {
     await withBindingLock(refId, async () => {
       const binding = repos.getBinding(db, refId);
-      if (!binding.health_check_enabled) return;
+      if (!binding.health_check_enabled && binding.lb_mode !== "weighted") return;
       const service = repos.getService(db, binding.service_id);
       if (!shouldPushDns(db, service)) return;
       const cnameTarget = binding.cname_target?.trim() || null;
       if (cnameTarget) return;
       const ips = repos.listServiceIps(db, service.id);
+      const poolIps = enabledServiceIps(db, service.id);
       const targetIps = repos.listBindingIps(db, binding.id);
       validateTargetIpsInPool(targetIps, ips);
-      const activeIps = computeActiveIps(db, "binding", refId);
-      const desiredIps = activeIps.length > 0 ? activeIps : targetIps;
+      const desiredIps = canApplyLb(poolIps, targetIps)
+        ? desiredAIps(db, "binding", refId, targetIps)
+        : targetIps;
       await syncBindingDns(
         db,
         cf,
@@ -1609,8 +1908,61 @@ export async function reconcileDnsForTarget(
   }
 
   const group = repos.getServiceGroup(db, refId);
-  if (!group.enabled || !group.domain?.trim() || !group.health_check_enabled) {
+  if (!group.enabled || !group.domain?.trim()) {
+    return;
+  }
+  if (!group.health_check_enabled && group.lb_mode !== "weighted") {
     return;
   }
   await syncGroupDomainDns(db, cf, refId);
 }
+
+export async function reconcileWeightedDns(
+  db: Db,
+  cf: CloudflareClient,
+): Promise<number> {
+  let n = 0;
+  for (const binding of repos.listAllBindings(db)) {
+    if (binding.lb_mode !== "weighted") continue;
+    if (binding.cname_target?.trim()) continue;
+    if (!isSharedPool(binding.target_ips ?? [])) continue;
+    try {
+      await withBindingLock(binding.id, async () => {
+        const latest = repos.getBinding(db, binding.id);
+        if (latest.lb_mode !== "weighted") return;
+        if (latest.cname_target?.trim()) return;
+        const service = repos.getService(db, latest.service_id);
+        if (!shouldPushDns(db, service)) return;
+        const targetIps = repos.listBindingIps(db, latest.id);
+        const poolIps = enabledServiceIps(db, service.id);
+        if (!canApplyLb(poolIps, targetIps)) return;
+        const ips = repos.listServiceIps(db, service.id);
+        validateTargetIpsInPool(targetIps, ips);
+        const desiredIps = desiredAIps(db, "binding", latest.id, targetIps);
+        await syncBindingDns(
+          db,
+          cf,
+          latest.id,
+          latest.domain_id,
+          latest.hostname,
+          desiredIps,
+          null,
+        );
+        n += 1;
+      });
+    } catch {
+      continue;
+    }
+  }
+  for (const group of repos.listServiceGroups(db)) {
+    if (group.lb_mode !== "weighted") continue;
+    if (!group.enabled || !group.domain?.trim()) continue;
+    try {
+      await syncGroupDomainDns(db, cf, group.id);
+      n += 1;
+    } catch {
+      continue;
+    }
+  }
+  return n;
+}

apps/api/src/services/vps-tracker-sync.ts

@@ -1,8 +1,53 @@
 import { resolve4 } from "node:dns/promises";
-import type { CfdmBindingSyncItem, ServiceBindingView } from "@cfdm/shared";
+import type { CfdmBindingSyncItem, LbMode, ServiceBindingView } from "@cfdm/shared";
 import { isIpLiteral } from "@cfdm/shared";
 import type { Db } from "@cfdm/db";
 import { repos, getAppSettingsSecrets, touchVpsTrackerSync } from "@cfdm/db";
+import { isSharedPool } from "./routing/pool.js";
+
+export function isLbMode(value: unknown): value is LbMode {
+  return value === "round_robin" || value === "failover" || value === "weighted";
+}
+
+/** lb_mode binding, иначе service group. */
+export function resolveLbModeForSync(
+  bindingLbMode: string | undefined | null,
+  groupLbMode?: string | null,
+): LbMode | undefined {
+  if (isLbMode(bindingLbMode)) return bindingLbMode;
+  if (isLbMode(groupLbMode)) return groupLbMode;
+  return undefined;
+}
+
+/** Effective HA only when the service has two or more unique origin IPs. */
+export function effectiveLbModeForSync(
+  bindingLbMode: string | undefined | null,
+  groupLbMode: string | undefined | null,
+  serviceIps: readonly string[],
+): LbMode | undefined {
+  if (!isSharedPool(serviceIps)) return undefined;
+  return resolveLbModeForSync(bindingLbMode, groupLbMode);
+}
+
+function groupLbModeForService(
+  db: Db,
+  serviceId: number,
+  cache: Map<number, LbMode | undefined>,
+): LbMode | undefined {
+  if (cache.has(serviceId)) return cache.get(serviceId);
+  let mode: LbMode | undefined;
+  try {
+    const service = repos.getService(db, serviceId);
+    if (service.service_group_id != null) {
+      const group = repos.getServiceGroup(db, service.service_group_id);
+      mode = resolveLbModeForSync(undefined, group.lb_mode);
+    }
+  } catch {
+    mode = undefined;
+  }
+  cache.set(serviceId, mode);
+  return mode;
+}
 
 function fqdnToDisplay(hostname: string, zoneName: string): string {
   if (hostname === "@" || !hostname.trim()) return zoneName;
@@ -126,10 +171,13 @@ export async function buildServiceSyncBindingsAsync(
   const allBindings = repos.listAllBindings(db);
   const index = buildBindingIndex(allBindings);
   const bindings = allBindings.filter((row) => row.service_id === serviceId);
+  const groupLbCache = new Map<number, LbMode | undefined>();
+  const groupLb = groupLbModeForService(db, serviceId, groupLbCache);
 
   const items: CfdmBindingSyncItem[] = [];
   for (const binding of bindings) {
     const ips = await resolveBindingIpsForSync(binding, serviceIps, index, db);
+    const lbMode = effectiveLbModeForSync(binding.lb_mode, groupLb, serviceIps);
     items.push({
       bindingId: binding.id,
       serviceId: service.id,
@@ -140,6 +188,7 @@ export async function buildServiceSyncBindingsAsync(
       hostname: binding.hostname,
       ips,
       cnameTarget: cnameTargetForSync(binding),
+      ...(lbMode ? { lbMode } : {}),
     });
   }
 
@@ -166,6 +215,7 @@ export async function buildAllSyncBindings(
   const bindings = repos.listAllBindings(db);
   const index = buildBindingIndex(bindings);
   const serviceIpCache = new Map<number, string[]>();
+  const groupLbCache = new Map<number, LbMode | undefined>();
 
   const items: CfdmBindingSyncItem[] = [];
   for (const binding of bindings) {
@@ -175,6 +225,8 @@ export async function buildAllSyncBindings(
       serviceIpCache.set(binding.service_id, serviceIps);
     }
     const ips = await resolveBindingIpsForSync(binding, serviceIps, index, db);
+    const groupLb = groupLbModeForService(db, binding.service_id, groupLbCache);
+    const lbMode = effectiveLbModeForSync(binding.lb_mode, groupLb, serviceIps);
     items.push({
       bindingId: binding.id,
       serviceId: binding.service_id,
@@ -185,6 +237,7 @@ export async function buildAllSyncBindings(
       hostname: binding.hostname,
       ips,
       cnameTarget: cnameTargetForSync(binding),
+      ...(lbMode ? { lbMode } : {}),
     });
   }
   return items;

apps/api/src/services/weighted-dns-scheduler.ts

@@ -0,0 +1,39 @@
+import type { FastifyInstance } from "fastify";
+import { AsyncTask, SimpleIntervalJob } from "toad-scheduler";
+import * as serviceConfigService from "./service-config-service.js";
+import { WEIGHTED_SLOT_MS } from "./routing/weighted.js";
+
+export const WEIGHTED_DNS_JOB_ID = "weighted-dns";
+
+export function createWeightedDnsTask(app: FastifyInstance): AsyncTask {
+  return new AsyncTask(
+    WEIGHTED_DNS_JOB_ID,
+    async () => {
+      const n = await serviceConfigService.reconcileWeightedDns(app.db, app.cf);
+      if (n > 0) {
+        app.log.info({ reconciled: n }, "weighted dns rotated");
+      }
+    },
+    (err) => {
+      app.log.warn({ err }, "weighted dns rotate failed");
+    },
+  );
+}
+
+export function scheduleWeightedDnsJob(
+  app: FastifyInstance,
+  task: AsyncTask,
+): void {
+  const scheduler = app.scheduler;
+  if (!scheduler) return;
+  if (scheduler.existsById(WEIGHTED_DNS_JOB_ID)) {
+    scheduler.removeById(WEIGHTED_DNS_JOB_ID);
+  }
+  scheduler.addSimpleIntervalJob(
+    new SimpleIntervalJob(
+      { seconds: WEIGHTED_SLOT_MS / 1000, runImmediately: true },
+      task,
+      { id: WEIGHTED_DNS_JOB_ID, preventOverrun: true },
+    ),
+  );
+}

apps/api/test/dns-identity.test.ts

@@ -0,0 +1,220 @@
+import { describe, expect, it } from "vitest";
+import { createMemoryDb, repos, runMigrations } from "@cfdm/db";
+import { SYNC_SYNCED } from "@cfdm/shared";
+import type { CloudflareClient } from "../src/lib/cf-client.js";
+import * as dnsService from "../src/services/dns-service.js";
+import { updateConfig } from "../src/services/service-config-service.js";
+
+type CfRec = {
+  id: string;
+  type: string;
+  name: string;
+  content: string;
+  ttl: number;
+  proxied: boolean;
+};
+
+function setupDb() {
+  const { db, sqlite } = createMemoryDb();
+  runMigrations(sqlite);
+  return db;
+}
+
+describe("DNS identity by name + content", () => {
+  it("deletes by name+IP when cached cf_record_id is stale/missing in CF", async () => {
+    const db = setupDb();
+    const remote: CfRec[] = [
+      {
+        id: "cf-live",
+        type: "A",
+        name: "nsgt.example.com",
+        content: "130.49.213.176",
+        ttl: 1,
+        proxied: false,
+      },
+    ];
+    const deleted: string[] = [];
+
+    const cf = {
+      listDnsRecords: async () => [...remote],
+      createDnsRecord: async () => {
+        throw new Error("create should not run");
+      },
+      updateDnsRecord: async () => {
+        throw new Error("update should not run");
+      },
+      deleteDnsRecord: async (_zoneId: string, id: string) => {
+        deleted.push(id);
+        const idx = remote.findIndex((r) => r.id === id);
+        if (idx >= 0) remote.splice(idx, 1);
+      },
+      verifyToken: async () => true,
+      listZones: async () => [],
+    } as unknown as CloudflareClient;
+
+    const domain = repos.createDomain(db, null, "example.com", "zone-1");
+    const local = repos.insertDnsRecord(
+      db,
+      domain.id,
+      "A",
+      "nsgt.example.com",
+      "130.49.213.176",
+      1,
+      false,
+      null,
+      SYNC_SYNCED,
+      "local",
+      "cf-stale-gone", // not present in Cloudflare
+    );
+
+    await dnsService.deleteRecord(db, cf, domain.id, local.id);
+
+    expect(deleted).toEqual(["cf-live"]);
+    expect(repos.listDnsByDomain(db, domain.id)).toEqual([]);
+    expect(remote).toEqual([]);
+  });
+
+  it("delete is no-op success when record already removed outside CFDM", async () => {
+    const db = setupDb();
+    const cf = {
+      listDnsRecords: async () => [],
+      deleteDnsRecord: async () => {
+        throw new Error("Record does not exist");
+      },
+      verifyToken: async () => true,
+      listZones: async () => [],
+    } as unknown as CloudflareClient;
+
+    const domain = repos.createDomain(db, null, "example.com", "zone-1");
+    const local = repos.insertDnsRecord(
+      db,
+      domain.id,
+      "A",
+      "nsgt.example.com",
+      "130.49.213.176",
+      1,
+      false,
+      null,
+      SYNC_SYNCED,
+      "local",
+      "cf-already-gone",
+    );
+
+    await expect(
+      dnsService.deleteRecord(db, cf, domain.id, local.id),
+    ).resolves.toBeUndefined();
+    expect(repos.listDnsByDomain(db, domain.id)).toEqual([]);
+  });
+
+  it("updateConfig can remove extra FQDN when CF id is stale", async () => {
+    const db = setupDb();
+    const remote: CfRec[] = [
+      {
+        id: "cf-gt",
+        type: "A",
+        name: "gt.example.com",
+        content: "130.49.213.176",
+        ttl: 1,
+        proxied: false,
+      },
+      {
+        id: "cf-nsgt-live",
+        type: "A",
+        name: "nsgt.example.com",
+        content: "130.49.213.176",
+        ttl: 1,
+        proxied: false,
+      },
+    ];
+    const deleted: string[] = [];
+
+    const cf = {
+      listDnsRecords: async () => [...remote],
+      createDnsRecord: async (
+        _zoneId: string,
+        payload: { type: string; name: string; content: string },
+      ) => {
+        const rec: CfRec = {
+          id: `cf-new-${remote.length}`,
+          type: payload.type,
+          name: payload.name,
+          content: payload.content,
+          ttl: 1,
+          proxied: false,
+        };
+        remote.push(rec);
+        return rec;
+      },
+      updateDnsRecord: async (
+        _zoneId: string,
+        id: string,
+        payload: { type: string; name: string; content: string },
+      ) => {
+        const idx = remote.findIndex((r) => r.id === id);
+        if (idx < 0) throw new Error("Record does not exist");
+        const rec: CfRec = {
+          id,
+          type: payload.type,
+          name: payload.name,
+          content: payload.content,
+          ttl: 1,
+          proxied: false,
+        };
+        remote[idx] = rec;
+        return rec;
+      },
+      deleteDnsRecord: async (_zoneId: string, id: string) => {
+        const idx = remote.findIndex((r) => r.id === id);
+        if (idx < 0) throw new Error("Record does not exist");
+        deleted.push(id);
+        remote.splice(idx, 1);
+      },
+      verifyToken: async () => true,
+      listZones: async () => [
+        { id: "zone-1", name: "example.com", status: "active" },
+      ],
+    } as unknown as CloudflareClient;
+
+    repos.createDomain(db, null, "example.com", "zone-1");
+    const service = repos.createService(db, "Main TG", "tg-pr");
+    repos.setServiceEnabled(db, service.id, true);
+
+    await updateConfig(db, cf, service.id, {
+      ips: ["130.49.213.176"],
+      domains: [
+        { fqdn: "gt.example.com", target_ips: ["130.49.213.176"] },
+        { fqdn: "nsgt.example.com", target_ips: ["130.49.213.176"] },
+      ],
+    });
+
+    // Poison cached id on extra binding's DNS row.
+    const nsgtBinding = repos
+      .listBindingsByService(db, service.id)
+      .find((b) => b.hostname === "nsgt")!;
+    const nsgtRecords = repos.listRecordsForBinding(db, nsgtBinding.id);
+    for (const row of nsgtRecords) {
+      repos.updateDnsFields(
+        db,
+        row.id,
+        row.record_type,
+        row.name,
+        row.content,
+        row.ttl,
+        row.proxied,
+        row.priority,
+        SYNC_SYNCED,
+        "cf-stale-nsgt",
+        null,
+      );
+    }
+
+    const view = await updateConfig(db, cf, service.id, {
+      ips: ["130.49.213.176"],
+      domains: [{ fqdn: "gt.example.com", target_ips: ["130.49.213.176"] }],
+    });
+
+    expect(view.domains.map((d) => d.fqdn)).toEqual(["gt.example.com"]);
+    expect(deleted).toContain("cf-nsgt-live");
+    expect(remote.some((r) => r.name.includes("nsgt"))).toBe(false);
+  });
+});

apps/api/test/dns-reconcile-extra.test.ts

@@ -0,0 +1,224 @@
+import { describe, expect, it } from "vitest";
+import { createMemoryDb, repos, runMigrations } from "@cfdm/db";
+import { SYNC_SYNCED } from "@cfdm/shared";
+import type { CloudflareClient } from "../src/lib/cf-client.js";
+import { updateConfig } from "../src/services/service-config-service.js";
+
+type CfRec = {
+  id: string;
+  type: string;
+  name: string;
+  content: string;
+  ttl: number;
+  proxied: boolean;
+};
+
+function setupDb() {
+  const { db, sqlite } = createMemoryDb();
+  runMigrations(sqlite);
+  return db;
+}
+
+describe("DNS reconcile for extra FQDN", () => {
+  it("publishes extra A even when local/CF already have CNAME on same name", async () => {
+    const db = setupDb();
+    const remote: CfRec[] = [
+      {
+        id: "cf-cname-nsgt",
+        type: "CNAME",
+        name: "nsgt.example.com",
+        content: "legacy.example.com",
+        ttl: 1,
+        proxied: false,
+      },
+    ];
+    const created: Array<{ type: string; name: string; content: string }> = [];
+    const deleted: string[] = [];
+
+    const cf = {
+      listDnsRecords: async () => remote,
+      createDnsRecord: async (
+        _zoneId: string,
+        payload: { type: string; name: string; content: string },
+      ) => {
+        created.push(payload);
+        const rec: CfRec = {
+          id: `cf-new-${created.length}`,
+          type: payload.type,
+          name: payload.name,
+          content: payload.content,
+          ttl: 1,
+          proxied: false,
+        };
+        remote.push(rec);
+        return rec;
+      },
+      updateDnsRecord: async (
+        _zoneId: string,
+        id: string,
+        payload: { type: string; name: string; content: string },
+      ) => {
+        const idx = remote.findIndex((r) => r.id === id);
+        const rec: CfRec = {
+          id,
+          type: payload.type,
+          name: payload.name,
+          content: payload.content,
+          ttl: 1,
+          proxied: false,
+        };
+        if (idx >= 0) remote[idx] = rec;
+        else remote.push(rec);
+        return rec;
+      },
+      deleteDnsRecord: async (_zoneId: string, id: string) => {
+        deleted.push(id);
+        const idx = remote.findIndex((r) => r.id === id);
+        if (idx >= 0) remote.splice(idx, 1);
+      },
+      verifyToken: async () => true,
+      listZones: async () => [
+        { id: "zone-1", name: "example.com", status: "active" },
+      ],
+    } as unknown as CloudflareClient;
+
+    const domain = repos.createDomain(db, null, "example.com", "zone-1");
+    // Leftover local CNAME (previous service / manual import).
+    repos.insertDnsRecord(
+      db,
+      domain.id,
+      "CNAME",
+      "nsgt.example.com",
+      "legacy.example.com",
+      1,
+      false,
+      null,
+      SYNC_SYNCED,
+      "cloudflare",
+      "cf-cname-nsgt",
+    );
+
+    const service = repos.createService(db, "MSK Hip", "tg-msk-hip");
+    repos.setServiceEnabled(db, service.id, true);
+
+    const view = await updateConfig(db, cf, service.id, {
+      ips: ["130.49.213.176"],
+      domains: [
+        {
+          fqdn: "gt.example.com",
+          target_ips: ["130.49.213.176"],
+        },
+        {
+          fqdn: "nsgt.example.com",
+          target_ips: ["130.49.213.176"],
+        },
+      ],
+    });
+
+    const nsgt = view.domains.find((d) => d.fqdn === "nsgt.example.com");
+    expect(nsgt?.record_type).toBe("A");
+    expect(nsgt?.target_ips).toEqual(["130.49.213.176"]);
+    expect(nsgt?.target_cname).toBeFalsy();
+
+    const binding = repos
+      .listBindingsByService(db, service.id)
+      .find((b) => b.hostname === "nsgt")!;
+    expect(repos.listBindingIps(db, binding.id)).toEqual(["130.49.213.176"]);
+    expect(deleted).toContain("cf-cname-nsgt");
+    expect(
+      created.some(
+        (r) =>
+          r.type === "A" &&
+          (r.name === "nsgt" || r.name === "nsgt.example.com") &&
+          r.content === "130.49.213.176",
+      ),
+    ).toBe(true);
+    expect(remote.some((r) => r.type === "CNAME" && r.name.includes("nsgt"))).toBe(
+      false,
+    );
+    expect(
+      remote.some(
+        (r) =>
+          r.type === "A" &&
+          r.name.includes("nsgt") &&
+          r.content === "130.49.213.176",
+      ),
+    ).toBe(true);
+  });
+
+  it("replaces stale A content for extra FQDN on the same hostname", async () => {
+    const db = setupDb();
+    const remote: CfRec[] = [
+      {
+        id: "cf-stale-a",
+        type: "A",
+        name: "nsgt.example.com",
+        content: "1.1.1.1",
+        ttl: 1,
+        proxied: false,
+      },
+    ];
+    const deleted: string[] = [];
+    const created: Array<{ type: string; name: string; content: string }> = [];
+
+    const cf = {
+      listDnsRecords: async () => [...remote],
+      createDnsRecord: async (
+        _zoneId: string,
+        payload: { type: string; name: string; content: string },
+      ) => {
+        created.push(payload);
+        const rec: CfRec = {
+          id: `cf-new-${created.length}`,
+          type: payload.type,
+          name: payload.name,
+          content: payload.content,
+          ttl: 1,
+          proxied: false,
+        };
+        remote.push(rec);
+        return rec;
+      },
+      updateDnsRecord: async (
+        _zoneId: string,
+        id: string,
+        payload: { type: string; name: string; content: string },
+      ) => ({
+        id,
+        type: payload.type,
+        name: payload.name,
+        content: payload.content,
+        ttl: 1,
+        proxied: false,
+      }),
+      deleteDnsRecord: async (_zoneId: string, id: string) => {
+        deleted.push(id);
+        const idx = remote.findIndex((r) => r.id === id);
+        if (idx >= 0) remote.splice(idx, 1);
+      },
+      verifyToken: async () => true,
+      listZones: async () => [
+        { id: "zone-1", name: "example.com", status: "active" },
+      ],
+    } as unknown as CloudflareClient;
+
+    repos.createDomain(db, null, "example.com", "zone-1");
+    const service = repos.createService(db, "MSK", "msk");
+    repos.setServiceEnabled(db, service.id, true);
+
+    await updateConfig(db, cf, service.id, {
+      ips: ["130.49.213.176"],
+      domains: [
+        { fqdn: "nsgt.example.com", target_ips: ["130.49.213.176"] },
+      ],
+    });
+
+    expect(deleted).toContain("cf-stale-a");
+    expect(
+      created.some(
+        (r) => r.type === "A" && r.content === "130.49.213.176",
+      ),
+    ).toBe(true);
+    expect(remote.map((r) => r.content)).toEqual(["130.49.213.176"]);
+  });
+});

apps/api/test/failover-log.test.ts

@@ -0,0 +1,112 @@
+import { describe, expect, it } from "vitest";
+import { createMemoryDb, repos, runMigrations } from "@cfdm/db";
+import { buildApp } from "../src/app.js";
+import { loadConfig } from "../src/config.js";
+import { failoverARecordDiff } from "../src/services/service-config-service.js";
+
+async function authHeaders(app: Awaited<ReturnType<typeof buildApp>>) {
+  const config = loadConfig();
+  const res = await app.inject({
+    method: "POST",
+    url: "/api/v1/auth/login",
+    payload: { username: config.adminUsername, password: "admin" },
+  });
+  expect(res.statusCode).toBe(200);
+  const { token } = res.json() as { token: string };
+  return { authorization: `Bearer ${token}` };
+}
+
+describe("failoverARecordDiff", () => {
+  it("diffs added and removed A contents", () => {
+    expect(
+      failoverARecordDiff(
+        ["130.49.213.153", "93.115.203.183"],
+        ["93.115.203.183"],
+      ),
+    ).toEqual({
+      added: [],
+      removed: ["130.49.213.153"],
+    });
+    expect(failoverARecordDiff(["10.0.0.1"], ["10.0.0.1", "10.0.0.2"])).toEqual({
+      added: ["10.0.0.2"],
+      removed: [],
+    });
+  });
+});
+
+describe("GET /services/:id/failover-log", () => {
+  it("returns add/remove rows for the service", async () => {
+    const app = await buildApp({
+      config: { ...loadConfig(), staticDir: null },
+      memory: true,
+    });
+    const headers = await authHeaders(app);
+
+    const domain = repos.createDomain(app.db, null, "rkns.top", "zone-1");
+    const service = repos.createService(app.db, "MSK Hip", "msk-hip");
+    const binding = repos.insertBinding(app.db, domain.id, service.id, "gt", null);
+
+    repos.insertFailoverLog(app.db, {
+      serviceId: service.id,
+      bindingId: binding.id,
+      fqdn: "gt.rkns.top",
+      entries: [
+        { ip: "130.49.213.153", action: "removed" },
+        { ip: "93.115.203.183", action: "added" },
+      ],
+    });
+
+    const res = await app.inject({
+      method: "GET",
+      url: `/api/v1/services/${service.id}/failover-log`,
+      headers,
+    });
+    expect(res.statusCode).toBe(200);
+    const body = res.json() as {
+      items: Array<{ ip: string; fqdn: string; action: string }>;
+    };
+    expect(body.items).toEqual(
+      expect.arrayContaining([
+        expect.objectContaining({
+          ip: "130.49.213.153",
+          fqdn: "gt.rkns.top",
+          action: "removed",
+        }),
+        expect.objectContaining({
+          ip: "93.115.203.183",
+          fqdn: "gt.rkns.top",
+          action: "added",
+        }),
+      ]),
+    );
+
+    await app.close();
+  });
+});
+
+describe("failover_log table", () => {
+  it("lists newest first", () => {
+    const { db, sqlite } = createMemoryDb();
+    runMigrations(sqlite);
+
+    const domain = repos.createDomain(db, null, "example.com", "zone-1");
+    const service = repos.createService(db, "Panel", "panel");
+    const binding = repos.insertBinding(db, domain.id, service.id, "panel", null);
+
+    repos.insertFailoverLog(db, {
+      serviceId: service.id,
+      bindingId: binding.id,
+      fqdn: "panel.example.com",
+      entries: [{ ip: "1.1.1.1", action: "removed" }],
+    });
+    repos.insertFailoverLog(db, {
+      serviceId: service.id,
+      bindingId: binding.id,
+      fqdn: "panel.example.com",
+      entries: [{ ip: "1.1.1.1", action: "added" }],
+    });
+
+    const rows = repos.listFailoverLogForService(db, service.id);
+    expect(rows.map((row) => row.action)).toEqual(["added", "removed"]);
+  });
+});

apps/api/test/health-check.test.ts

@@ -291,6 +291,77 @@ describe("health-check state derivation via runAllChecks", () => {
     expect(targets[0]?.ip).toBe("2.59.161.102");
     expect(targets[0]?.hostname).toBe("s.rkns.top");
   });
+
+  it("does not mark node unhealthy when binding majority is OK and group local fails", async () => {
+    const { createMemoryDb, repos, runMigrations } = await import("@cfdm/db");
+    const { db, sqlite } = createMemoryDb();
+    runMigrations(sqlite);
+
+    const tcp = await startTcpServer();
+    try {
+      const domain = repos.createDomain(db, null, "example.com", "zone-id");
+      const group = repos.createServiceGroup(
+        db,
+        "VPN",
+        "vpn",
+        null,
+        "vpn.example.com",
+        {
+          health_check_enabled: true,
+          health_check_type: "http",
+          health_check_port: 1,
+          health_check_timeout_ms: 200,
+          health_check_path: "/",
+        },
+      );
+      const service = repos.createService(db, "Svc", "svc");
+      repos.setServiceGroup(db, service.id, group.id);
+      repos.setServiceEnabled(db, service.id, true);
+      const binding = repos.insertBinding(db, domain.id, service.id, "@", null);
+      repos.updateBindingLbConfig(db, binding.id, {
+        health_check_enabled: true,
+        health_check_type: "tcp",
+        health_check_port: tcp.port,
+        health_check_timeout_ms: 500,
+      });
+      repos.replaceBindingIpsWithMeta(db, binding.id, [
+        { ip: "127.0.0.1", weight: 1, priority: 1 },
+      ]);
+      const node = repos.findNodeByIp(db, "127.0.0.1");
+      expect(node).not.toBeNull();
+
+      await healthCheckService.runAllChecks(db, {
+        probeGapMs: 0,
+        thresholds: {
+          degradedFailures: 1,
+          downFailures: 1,
+          latencyWarnMs: 1000,
+        },
+      });
+
+      const bindingHealth = repos.getIpHealthStatusRow(
+        db,
+        "binding",
+        binding.id,
+        "127.0.0.1",
+      );
+      const groupHealth = repos.getIpHealthStatusRow(
+        db,
+        "group",
+        group.id,
+        "127.0.0.1",
+      );
+      const after = repos.getNode(db, node!.id);
+
+      expect(bindingHealth?.status).toBe("up");
+      expect(groupHealth?.status).toBe("down");
+      expect(after.health_status).toBe("healthy");
+      expect(after.consecutive_failures).toBe(0);
+      expect(after.last_failure_reason).toBeNull();
+    } finally {
+      await new Promise<void>((resolve) => tcp.server.close(() => resolve()));
+    }
+  });
 });
 
 describe("CNAME health mapped onto service IPs", () => {
@@ -305,6 +376,7 @@ describe("CNAME health mapped onto service IPs", () => {
     repos.replaceServiceIps(db, service.id, ["2.59.161.102"]);
     const binding = repos.insertBinding(db, domain.id, service.id, "s", null);
     repos.setBindingCnameTarget(db, binding.id, "ihome.rkns.top");
+    repos.updateBindingLbConfig(db, binding.id, { health_check_enabled: true });
     repos.upsertIpHealthStatus(
       db,
       "binding",
@@ -326,4 +398,157 @@ describe("CNAME health mapped onto service IPs", () => {
       }),
     ]);
   });
+
+  it("getView is up when any binding IP is up", async () => {
+    const { createMemoryDb, repos, runMigrations } = await import("@cfdm/db");
+    const { getView } = await import("../src/services/service-config-service.js");
+    const { db, sqlite } = createMemoryDb();
+    runMigrations(sqlite);
+
+    const domain = repos.createDomain(db, null, "rkns.top", "zone-id");
+    const service = repos.createService(db, "MSK Hip", "msk-hip");
+    repos.replaceServiceIps(db, service.id, ["10.0.0.1", "10.0.0.2"]);
+    const binding = repos.insertBinding(db, domain.id, service.id, "gt", null);
+    repos.replaceBindingIpsWithMeta(db, binding.id, [
+      { ip: "10.0.0.1", weight: 1, priority: 1 },
+      { ip: "10.0.0.2", weight: 1, priority: 1 },
+    ]);
+    repos.updateBindingLbConfig(db, binding.id, { health_check_enabled: true });
+    repos.upsertIpHealthStatus(
+      db,
+      "binding",
+      binding.id,
+      "10.0.0.1",
+      "up",
+      12,
+      0,
+      null,
+    );
+    repos.upsertIpHealthStatus(
+      db,
+      "binding",
+      binding.id,
+      "10.0.0.2",
+      "down",
+      null,
+      5,
+      "timeout",
+    );
+
+    const view = await getView(db, service.id);
+    expect(view.health_status).toBe("up");
+  });
+
+  it("getView shows live OK over hysteresis unknown", async () => {
+    const { createMemoryDb, repos, runMigrations } = await import("@cfdm/db");
+    const { getView } = await import("../src/services/service-config-service.js");
+    const { db, sqlite } = createMemoryDb();
+    runMigrations(sqlite);
+
+    const domain = repos.createDomain(db, null, "rkns.top", "zone-id");
+    const service = repos.createService(db, "RW Panel", "rw-panel");
+    repos.replaceServiceIps(db, service.id, ["2.59.161.102"]);
+    const binding = repos.insertBinding(db, domain.id, service.id, "c", null);
+    repos.replaceBindingIpsWithMeta(db, binding.id, [
+      { ip: "2.59.161.102", weight: 1, priority: 1 },
+    ]);
+    repos.updateBindingLbConfig(db, binding.id, { health_check_enabled: true });
+    repos.upsertIpHealthStatus(
+      db,
+      "binding",
+      binding.id,
+      "2.59.161.102",
+      "unknown",
+      63,
+      0,
+      null,
+      1,
+    );
+    repos.insertHealthProbeLog(db, {
+      scope: "binding",
+      refId: binding.id,
+      ip: "2.59.161.102",
+      provider: "local",
+      status: "up",
+      ok: true,
+      latencyMs: 63,
+      colo: null,
+      error: null,
+    });
+
+    const view = await getView(db, service.id);
+    expect(view.health_status).toBe("up");
+    expect(view.ip_health[0]?.status).toBe("up");
+    expect(view.ip_health[0]?.latency_ms).toBe(63);
+  });
+
+  it("getView masks stale down when health-check is disabled", async () => {
+    const { createMemoryDb, repos, runMigrations } = await import("@cfdm/db");
+    const { getView } = await import("../src/services/service-config-service.js");
+    const { db, sqlite } = createMemoryDb();
+    runMigrations(sqlite);
+
+    const domain = repos.createDomain(db, null, "rkns.top", "zone-id");
+    const service = repos.createService(db, "Main TG", "main-tg");
+    repos.setServiceEnabled(db, service.id, true);
+    repos.replaceServiceIps(db, service.id, ["130.49.213.176"]);
+    const binding = repos.insertBinding(db, domain.id, service.id, "gt", null);
+    repos.replaceBindingIpsWithMeta(db, binding.id, [
+      { ip: "130.49.213.176", weight: 1, priority: 1 },
+    ]);
+    repos.updateBindingLbConfig(db, binding.id, { health_check_enabled: false });
+    repos.upsertIpHealthStatus(
+      db,
+      "binding",
+      binding.id,
+      "130.49.213.176",
+      "down",
+      null,
+      5,
+      "timeout",
+    );
+
+    const view = await getView(db, service.id);
+    expect(view.health_status).toBe("unknown");
+    expect(view.ip_health).toEqual([
+      expect.objectContaining({
+        ip: "130.49.213.176",
+        status: "unknown",
+        latency_ms: null,
+        last_error: null,
+      }),
+    ]);
+  });
+
+  it("getView masks stale down when IP is disabled in pool", async () => {
+    const { createMemoryDb, repos, runMigrations } = await import("@cfdm/db");
+    const { getView } = await import("../src/services/service-config-service.js");
+    const { db, sqlite } = createMemoryDb();
+    runMigrations(sqlite);
+
+    const domain = repos.createDomain(db, null, "rkns.top", "zone-id");
+    const service = repos.createService(db, "Main TG", "main-tg");
+    repos.setServiceEnabled(db, service.id, true);
+    repos.replaceServiceIps(db, service.id, ["130.49.213.176"]);
+    repos.setServiceIpEnabled(db, service.id, "130.49.213.176", false);
+    const binding = repos.insertBinding(db, domain.id, service.id, "gt", null);
+    repos.replaceBindingIpsWithMeta(db, binding.id, [
+      { ip: "130.49.213.176", weight: 1, priority: 1 },
+    ]);
+    repos.updateBindingLbConfig(db, binding.id, { health_check_enabled: true });
+    repos.upsertIpHealthStatus(
+      db,
+      "binding",
+      binding.id,
+      "130.49.213.176",
+      "down",
+      null,
+      5,
+      "timeout",
+    );
+
+    const view = await getView(db, service.id);
+    expect(view.health_status).toBe("unknown");
+    expect(view.ip_health[0]?.status).toBe("unknown");
+  });
 });

apps/api/test/lb-reconcile.test.ts

@@ -1,9 +1,12 @@
 import { describe, expect, it } from "vitest";
 import {
+  resolveDesiredAIps,
   selectActiveIpsByMode,
+  shouldRecordFailoverDnsDiff,
   type LbIpRow,
   type LbTargetConfig,
 } from "../src/services/service-config-service.js";
+import { WEIGHTED_SLOT_MS } from "../src/services/routing/weighted.js";
 
 function row(
   ip: string,
@@ -17,6 +20,11 @@ function row(
   };
 }
 
+const weightedConfig: LbTargetConfig = {
+  lb_mode: "weighted",
+  health_check_enabled: true,
+};
+
 describe("selectActiveIpsByMode", () => {
   it("round_robin returns all healthy ips, falls back to all if none healthy", () => {
     const config: LbTargetConfig = {
@@ -62,6 +70,18 @@ describe("selectActiveIpsByMode", () => {
     expect(selectActiveIpsByMode(config, rows)).toEqual(["1.1.1.1"]);
   });
 
+  it("failover returns recovering unknown primary immediately", () => {
+    const config: LbTargetConfig = {
+      lb_mode: "failover",
+      health_check_enabled: true,
+    };
+    const rows = [
+      row("1.1.1.1", { priority: 1, health: "unknown" }),
+      row("2.2.2.2", { priority: 2, health: "up" }),
+    ];
+    expect(selectActiveIpsByMode(config, rows)).toEqual(["1.1.1.1"]);
+  });
+
   it("failover falls back to min-priority ip among all when none healthy", () => {
     const config: LbTargetConfig = {
       lb_mode: "failover",
@@ -75,23 +95,61 @@ describe("selectActiveIpsByMode", () => {
     expect(selectActiveIpsByMode(config, rows)).toEqual(["2.2.2.2"]);
   });
 
-  it("weighted returns all healthy ips (one A per ip; weights stored for display)", () => {
-    const config: LbTargetConfig = {
-      lb_mode: "weighted",
-      health_check_enabled: true,
-    };
+  it("weighted 1:3 picks the lighter ip on slot 0 and the heavier on slot 1", () => {
     const rows = [
-      row("1.1.1.1", { weight: 3, health: "up" }),
-      row("2.2.2.2", { weight: 1, health: "up" }),
-      row("3.3.3.3", { weight: 2, health: "down" }),
+      row("1.1.1.1", { weight: 1, health: "up" }),
+      row("2.2.2.2", { weight: 3, health: "up" }),
     ];
-    expect(selectActiveIpsByMode(config, rows).sort()).toEqual([
-      "1.1.1.1",
+    expect(selectActiveIpsByMode(weightedConfig, rows, 0)).toEqual(["1.1.1.1"]);
+    expect(selectActiveIpsByMode(weightedConfig, rows, WEIGHTED_SLOT_MS)).toEqual([
       "2.2.2.2",
     ]);
   });
 
-  it("round_robin excludes unknown when another ip is up", () => {
+  it("weighted excludes down ips from the cycle", () => {
+    const rows = [
+      row("1.1.1.1", { weight: 1, health: "up" }),
+      row("2.2.2.2", { weight: 3, health: "down" }),
+    ];
+    expect(selectActiveIpsByMode(weightedConfig, rows, 0)).toEqual(["1.1.1.1"]);
+    expect(
+      selectActiveIpsByMode(weightedConfig, rows, WEIGHTED_SLOT_MS),
+    ).toEqual(["1.1.1.1"]);
+  });
+
+  it("weighted includes recovering unknown in the cycle", () => {
+    const rows = [
+      row("1.1.1.1", { weight: 1, health: "up" }),
+      row("2.2.2.2", { weight: 3, health: "unknown" }),
+    ];
+    expect(selectActiveIpsByMode(weightedConfig, rows, 0)).toEqual(["1.1.1.1"]);
+    expect(
+      selectActiveIpsByMode(weightedConfig, rows, WEIGHTED_SLOT_MS),
+    ).toEqual(["2.2.2.2"]);
+  });
+
+  it("weighted with one ip always returns that ip", () => {
+    expect(
+      selectActiveIpsByMode(weightedConfig, [row("1.1.1.1", { weight: 5 })], 0),
+    ).toEqual(["1.1.1.1"]);
+  });
+
+  it("weighted with all unknown rotates across every ip", () => {
+    const rows = [
+      row("1.1.1.1", { weight: 1, health: "unknown" }),
+      row("2.2.2.2", { weight: 3, health: "unknown" }),
+    ];
+    expect(selectActiveIpsByMode(weightedConfig, rows, 0)).toEqual(["1.1.1.1"]);
+    expect(selectActiveIpsByMode(weightedConfig, rows, WEIGHTED_SLOT_MS)).toEqual([
+      "2.2.2.2",
+    ]);
+  });
+
+  it("weighted returns empty array for no rows", () => {
+    expect(selectActiveIpsByMode(weightedConfig, [], 0)).toEqual([]);
+  });
+
+  it("round_robin puts recovering unknown back with live ips", () => {
     const config: LbTargetConfig = {
       lb_mode: "round_robin",
       health_check_enabled: true,
@@ -100,7 +158,25 @@ describe("selectActiveIpsByMode", () => {
       row("1.1.1.1", { health: "up" }),
       row("2.2.2.2", { health: "unknown" }),
     ];
-    expect(selectActiveIpsByMode(config, rows)).toEqual(["1.1.1.1"]);
+    expect(selectActiveIpsByMode(config, rows).sort()).toEqual([
+      "1.1.1.1",
+      "2.2.2.2",
+    ]);
+  });
+
+  it("round_robin keeps degraded in the pool with live ips", () => {
+    const config: LbTargetConfig = {
+      lb_mode: "round_robin",
+      health_check_enabled: true,
+    };
+    const rows = [
+      row("1.1.1.1", { health: "up" }),
+      row("2.2.2.2", { health: "degraded" }),
+    ];
+    expect(selectActiveIpsByMode(config, rows).sort()).toEqual([
+      "1.1.1.1",
+      "2.2.2.2",
+    ]);
   });
 
   it("returns empty array for no rows", () => {
@@ -111,3 +187,115 @@ describe("selectActiveIpsByMode", () => {
     expect(selectActiveIpsByMode(config, [])).toEqual([]);
   });
 });
+
+describe("resolveDesiredAIps", () => {
+  it("keeps a dedicated single IP even when down", () => {
+    expect(
+      resolveDesiredAIps(
+        weightedConfig,
+        [row("1.1.1.1", { health: "down" })],
+        ["1.1.1.1"],
+      ),
+    ).toEqual(["1.1.1.1"]);
+  });
+
+  it("does not drain when the service has only one unique IP", () => {
+    expect(
+      resolveDesiredAIps(
+        weightedConfig,
+        [row("1.1.1.1", { health: "down" })],
+        ["1.1.1.1", "1.1.1.1"],
+        0,
+        ["1.1.1.1"],
+      ),
+    ).toEqual(["1.1.1.1", "1.1.1.1"]);
+  });
+
+  it("does not apply overlay when service pool is a single IP", () => {
+    const rows = [
+      row("1.1.1.1", { health: "up" }),
+      row("2.2.2.2", { health: "down" }),
+    ];
+    expect(
+      resolveDesiredAIps(
+        weightedConfig,
+        rows,
+        ["1.1.1.1", "2.2.2.2"],
+        0,
+        ["1.1.1.1"],
+      ),
+    ).toEqual(["1.1.1.1", "2.2.2.2"]);
+  });
+
+  it("applies weighted overlay on a shared pool", () => {
+    const rows = [
+      row("1.1.1.1", { weight: 1, health: "up" }),
+      row("2.2.2.2", { weight: 3, health: "up" }),
+    ];
+    expect(
+      resolveDesiredAIps(weightedConfig, rows, ["1.1.1.1", "2.2.2.2"], 0),
+    ).toEqual(selectActiveIpsByMode(weightedConfig, rows, 0));
+  });
+});
+
+describe("shouldRecordFailoverDnsDiff", () => {
+  it("skips duplicate listings of the same IP", () => {
+    expect(
+      shouldRecordFailoverDnsDiff({
+        configuredIps: ["1.1.1.1", "1.1.1.1"],
+        lbMode: "failover",
+        added: [],
+        removed: ["1.1.1.1"],
+        downIps: new Set(["1.1.1.1"]),
+      }),
+    ).toBe(false);
+  });
+
+  it("skips dedicated extra-FQDN", () => {
+    expect(
+      shouldRecordFailoverDnsDiff({
+        configuredIps: ["1.1.1.1"],
+        lbMode: "failover",
+        added: [],
+        removed: ["1.1.1.1"],
+        downIps: new Set(["1.1.1.1"]),
+      }),
+    ).toBe(false);
+  });
+
+  it("skips weighted live-to-live slot swap", () => {
+    expect(
+      shouldRecordFailoverDnsDiff({
+        configuredIps: ["1.1.1.1", "2.2.2.2"],
+        lbMode: "weighted",
+        added: ["2.2.2.2"],
+        removed: ["1.1.1.1"],
+        downIps: new Set(),
+      }),
+    ).toBe(false);
+  });
+
+  it("logs weighted swap when a down ip leaves the pool", () => {
+    expect(
+      shouldRecordFailoverDnsDiff({
+        configuredIps: ["1.1.1.1", "2.2.2.2"],
+        lbMode: "weighted",
+        added: ["2.2.2.2"],
+        removed: ["1.1.1.1"],
+        downIps: new Set(["1.1.1.1"]),
+      }),
+    ).toBe(true);
+  });
+
+  it("logs failover diffs on a shared pool", () => {
+    expect(
+      shouldRecordFailoverDnsDiff({
+        configuredIps: ["1.1.1.1", "2.2.2.2"],
+        lbMode: "failover",
+        added: ["2.2.2.2"],
+        removed: ["1.1.1.1"],
+        downIps: new Set(["1.1.1.1"]),
+      }),
+    ).toBe(true);
+  });
+});

apps/api/test/service-bindings-prune.test.ts

@@ -78,4 +78,53 @@ describe("service bindings prune", () => {
     expect(bindings).toHaveLength(2);
     expect(bindings.map((b) => b.hostname).sort()).toEqual(["api", "www"]);
   });
+
+  it("updateConfig with extra FQDN per IP does not throw when group health-check is on", async () => {
+    const db = setupDb();
+    const cf = mockCf();
+    const health = {
+      health_check_enabled: true,
+      health_check_type: "tcp" as const,
+      health_check_port: 443,
+      health_check_providers: ["local", "cloudflare", "globalping"] as const,
+      health_check_aggregate: "majority" as const,
+    };
+
+    repos.createDomain(db, null, "example.com", "cf-zone-example");
+    const group = repos.createServiceGroup(
+      db,
+      "VPN",
+      "vpn",
+      null,
+      "vpn.example.com",
+      { ...health },
+    );
+    const service = repos.createService(db, "GT", "gt");
+    repos.setServiceGroup(db, service.id, group.id);
+    repos.setServiceEnabled(db, service.id, true);
+
+    const view = await updateConfig(db, cf, service.id, {
+      ips: ["93.115.203.183", "130.49.213.153"],
+      domains: [
+        {
+          fqdn: "gt.example.com",
+          target_ips: ["93.115.203.183", "130.49.213.153"],
+          ...health,
+        },
+        {
+          fqdn: "rutg.example.com",
+          target_ips: ["93.115.203.183"],
+          ...health,
+        },
+        {
+          fqdn: "nsgt.example.com",
+          target_ips: ["130.49.213.153"],
+          ...health,
+        },
+      ],
+    });
+
+    expect(view.domains).toHaveLength(3);
+    expect(view.ips.sort()).toEqual(["130.49.213.153", "93.115.203.183"]);
+  });
 });

apps/api/test/services-create-list.test.ts

@@ -6,6 +6,7 @@ import { buildApp } from "../src/app.js";
 import { loadConfig } from "../src/config.js";
 import {
   listGroupViews,
+  toggleServiceIp,
   updateConfig,
 } from "../src/services/service-config-service.js";
 
@@ -250,7 +251,7 @@ describe("create service then list groups", () => {
     await app.close();
   });
 
-  it("PATCH /services/:id/ips/toggle keeps IP in pool and removes it from A-binding", async () => {
+  it("PATCH /services/:id/ips/toggle keeps IP in pool and in A-binding", async () => {
     const app = await buildApp({
       config: { ...loadConfig(), staticDir: null },
       memory: true,
@@ -280,6 +281,18 @@ describe("create service then list groups", () => {
     expect(createRes.statusCode).toBe(200);
     const created = createRes.json() as { id: number };
 
+    const domainPayload = {
+      lb_mode: "round_robin" as const,
+      health_check_enabled: false,
+      health_check_type: "tcp" as const,
+      health_check_port: 443,
+      health_check_path: null,
+      health_check_expected_status: null,
+      health_check_interval_sec: 30,
+      health_check_timeout_ms: 3000,
+      health_check_verify_tls: false,
+    };
+
     await updateConfig(app.db, cf, created.id, {
       ips: ["1.2.3.4", "5.6.7.8"],
       service_group_id: group.id,
@@ -289,23 +302,81 @@ describe("create service then list groups", () => {
           target_ips: ["1.2.3.4", "5.6.7.8"],
           target_ip_weights: { "1.2.3.4": 1, "5.6.7.8": 1 },
           target_ip_priorities: { "1.2.3.4": 1, "5.6.7.8": 1 },
-          lb_mode: "round_robin",
-          health_check_enabled: false,
-          health_check_type: "tcp",
-          health_check_port: 443,
-          health_check_path: null,
-          health_check_expected_status: null,
-          health_check_interval_sec: 30,
-          health_check_timeout_ms: 3000,
-          health_check_verify_tls: false,
+          ...domainPayload,
+        },
+        {
+          fqdn: "extra.example.com",
+          target_ips: ["1.2.3.4"],
+          target_ip_weights: { "1.2.3.4": 1 },
+          target_ip_priorities: { "1.2.3.4": 1 },
+          ...domainPayload,
         },
       ],
     });
 
-    // HTTP toggle uses request.server.cf; disable DNS push so the test
-    // does not call the real Cloudflare client.
-    repos.setServiceEnabled(app.db, created.id, false);
+    const commonBinding = repos
+      .listBindingsByService(app.db, created.id)
+      .find((b) => b.hostname === "panel")!;
+    const extraBinding = repos
+      .listBindingsByService(app.db, created.id)
+      .find((b) => b.hostname === "extra")!;
+    expect(repos.listBindingIps(app.db, commonBinding.id)).toEqual(
+      expect.arrayContaining(["1.2.3.4", "5.6.7.8"]),
+    );
+    expect(
+      repos
+        .listRecordsForBinding(app.db, commonBinding.id)
+        .map((r) => r.content)
+        .sort(),
+    ).toEqual(["1.2.3.4", "5.6.7.8"]);
 
+    // Direct service call with mock CF — keep HTTP path free of real Cloudflare.
+    await toggleServiceIp(app.db, cf, created.id, "1.2.3.4", false);
+
+    expect(repos.listServiceIps(app.db, created.id)).toEqual(
+      expect.arrayContaining(["1.2.3.4", "5.6.7.8"]),
+    );
+    expect(
+      repos.listServiceIpRows(app.db, created.id).find((r) => r.ip === "1.2.3.4")
+        ?.enabled,
+    ).toBe(false);
+    // Common + per-IP bindings keep configured IPs (UI hydrate stays stable).
+    expect(repos.listBindingIps(app.db, commonBinding.id)).toEqual(
+      expect.arrayContaining(["1.2.3.4", "5.6.7.8"]),
+    );
+    expect(repos.listBindingIps(app.db, extraBinding.id)).toEqual(["1.2.3.4"]);
+    // DNS for common FQDN drops the disabled IP only.
+    expect(
+      repos
+        .listRecordsForBinding(app.db, commonBinding.id)
+        .map((r) => r.content)
+        .sort(),
+    ).toEqual(["5.6.7.8"]);
+    // Per-IP extra FQDN has no enabled targets → A records removed.
+    expect(repos.listRecordsForBinding(app.db, extraBinding.id)).toEqual([]);
+
+    await toggleServiceIp(app.db, cf, created.id, "1.2.3.4", true);
+    expect(
+      repos.listServiceIpRows(app.db, created.id).find((r) => r.ip === "1.2.3.4")
+        ?.enabled,
+    ).toBe(true);
+    expect(repos.listBindingIps(app.db, commonBinding.id)).toEqual(
+      expect.arrayContaining(["1.2.3.4", "5.6.7.8"]),
+    );
+    expect(
+      repos
+        .listRecordsForBinding(app.db, commonBinding.id)
+        .map((r) => r.content)
+        .sort(),
+    ).toEqual(["1.2.3.4", "5.6.7.8"]);
+    expect(
+      repos
+        .listRecordsForBinding(app.db, extraBinding.id)
+        .map((r) => r.content),
+    ).toEqual(["1.2.3.4"]);
+
+    // HTTP toggle still updates ip_enabled without mutating bindings.
+    repos.setServiceEnabled(app.db, created.id, false);
     const offRes = await app.inject({
       method: "PATCH",
       url: `/api/v1/services/${created.id}/ips/toggle`,
@@ -319,24 +390,117 @@ describe("create service then list groups", () => {
     };
     expect(offView.ips).toEqual(expect.arrayContaining(["1.2.3.4", "5.6.7.8"]));
     expect(offView.ip_enabled["1.2.3.4"]).toBe(false);
-    expect(offView.ip_enabled["5.6.7.8"]).toBe(true);
-
-    const binding = repos.listBindingsByService(app.db, created.id)[0]!;
-    expect(repos.listBindingIps(app.db, binding.id)).toEqual(["5.6.7.8"]);
-
-    const onRes = await app.inject({
-      method: "PATCH",
-      url: `/api/v1/services/${created.id}/ips/toggle`,
-      headers,
-      payload: { ip: "1.2.3.4", enabled: true },
-    });
-    expect(onRes.statusCode).toBe(200);
-    const onView = onRes.json() as { ip_enabled: Record<string, boolean> };
-    expect(onView.ip_enabled["1.2.3.4"]).toBe(true);
-    expect(repos.listBindingIps(app.db, binding.id)).toEqual(
+    expect(repos.listBindingIps(app.db, commonBinding.id)).toEqual(
       expect.arrayContaining(["1.2.3.4", "5.6.7.8"]),
     );
 
     await app.close();
   });
+
+  it("GET /services/:id repairs multi-IP bindings shrunk below the pool", async () => {
+    const app = await buildApp({
+      config: { ...loadConfig(), staticDir: null },
+      memory: true,
+    });
+    const headers = await authHeaders(app);
+    const cf = mockCf();
+
+    repos.createDomain(app.db, null, "example.com", "zone-1");
+    const group = repos.createServiceGroup(
+      app.db,
+      "VPN",
+      "vpn-repair",
+      null,
+      "vpn.example.com",
+    );
+
+    const createRes = await app.inject({
+      method: "POST",
+      url: "/api/v1/services",
+      headers,
+      payload: {
+        name: "Repair",
+        slug: "panel-ip-repair",
+        service_group_id: group.id,
+      },
+    });
+    expect(createRes.statusCode).toBe(200);
+    const created = createRes.json() as { id: number };
+
+    await updateConfig(app.db, cf, created.id, {
+      ips: ["1.2.3.4", "5.6.7.8", "9.9.9.9"],
+      service_group_id: group.id,
+      domains: [
+        {
+          fqdn: "gw.example.com",
+          target_ips: ["1.2.3.4", "5.6.7.8", "9.9.9.9"],
+          target_ip_weights: { "1.2.3.4": 1, "5.6.7.8": 1, "9.9.9.9": 1 },
+          target_ip_priorities: { "1.2.3.4": 1, "5.6.7.8": 1, "9.9.9.9": 1 },
+          lb_mode: "round_robin",
+          health_check_enabled: false,
+          health_check_type: "tcp",
+          health_check_port: 443,
+          health_check_path: null,
+          health_check_expected_status: null,
+          health_check_interval_sec: 30,
+          health_check_timeout_ms: 3000,
+          health_check_verify_tls: false,
+        },
+        {
+          fqdn: "extra.example.com",
+          target_ips: ["1.2.3.4"],
+          target_ip_weights: { "1.2.3.4": 1 },
+          target_ip_priorities: { "1.2.3.4": 1 },
+          lb_mode: "round_robin",
+          health_check_enabled: false,
+          health_check_type: "tcp",
+          health_check_port: 443,
+          health_check_path: null,
+          health_check_expected_status: null,
+          health_check_interval_sec: 30,
+          health_check_timeout_ms: 3000,
+          health_check_verify_tls: false,
+        },
+      ],
+    });
+
+    const commonBinding = repos
+      .listBindingsByService(app.db, created.id)
+      .find((b) => b.hostname === "gw")!;
+    const extraBinding = repos
+      .listBindingsByService(app.db, created.id)
+      .find((b) => b.hostname === "extra")!;
+
+    // Simulate legacy toggle damage: shrink common binding, leave extra alone.
+    repos.replaceBindingIpsWithMeta(app.db, commonBinding.id, [
+      { ip: "1.2.3.4", weight: 1, priority: 1 },
+      { ip: "5.6.7.8", weight: 1, priority: 1 },
+    ]);
+    expect(repos.listBindingIps(app.db, commonBinding.id)).toEqual([
+      "1.2.3.4",
+      "5.6.7.8",
+    ]);
+
+    const getRes = await app.inject({
+      method: "GET",
+      url: `/api/v1/services/${created.id}`,
+      headers,
+    });
+    expect(getRes.statusCode).toBe(200);
+    const view = getRes.json() as {
+      domains: Array<{ fqdn: string; target_ips: string[] }>;
+    };
+    const gw = view.domains.find((d) => d.fqdn === "gw.example.com");
+    const extra = view.domains.find((d) => d.fqdn === "extra.example.com");
+    expect(gw?.target_ips.sort()).toEqual(["1.2.3.4", "5.6.7.8", "9.9.9.9"]);
+    expect(extra?.target_ips).toEqual(["1.2.3.4"]);
+    expect(repos.listBindingIps(app.db, commonBinding.id).sort()).toEqual([
+      "1.2.3.4",
+      "5.6.7.8",
+      "9.9.9.9",
+    ]);
+    expect(repos.listBindingIps(app.db, extraBinding.id)).toEqual(["1.2.3.4"]);
+
+    await app.close();
+  });
 });

apps/api/test/vps-tracker-sync.test.ts

@@ -1,6 +1,13 @@
 import { describe, expect, it } from "vitest";
-import type { ServiceBindingView } from "@cfdm/shared";
-import { resolveBindingIpsForSync } from "../src/services/vps-tracker-sync.js";
+import {
+  cfdmBindingSyncItemSchema,
+  type ServiceBindingView,
+} from "@cfdm/shared";
+import {
+  resolveBindingIpsForSync,
+  resolveLbModeForSync,
+  effectiveLbModeForSync,
+} from "../src/services/vps-tracker-sync.js";
 
 function binding(
   partial: Partial<ServiceBindingView> &
@@ -160,3 +167,71 @@ describe("resolveBindingIpsForSync", () => {
     expect(ips).toEqual(["203.0.113.55"]);
   });
 });
+
+describe("resolveLbModeForSync", () => {
+  it("prefers binding lb_mode", () => {
+    expect(resolveLbModeForSync("failover", "round_robin")).toBe("failover");
+  });
+
+  it("falls back to service group lb_mode", () => {
+    expect(resolveLbModeForSync("off", "weighted")).toBe("weighted");
+    expect(resolveLbModeForSync(undefined, "round_robin")).toBe("round_robin");
+  });
+
+  it("returns undefined when neither is a known mode", () => {
+    expect(resolveLbModeForSync("off", "off")).toBeUndefined();
+    expect(resolveLbModeForSync(null, undefined)).toBeUndefined();
+  });
+});
+
+describe("effectiveLbModeForSync", () => {
+  it("omits mode when unique origin IPs are below two", () => {
+    expect(
+      effectiveLbModeForSync("round_robin", "failover", ["203.0.113.10"]),
+    ).toBeUndefined();
+    expect(
+      effectiveLbModeForSync("round_robin", "failover", [
+        "203.0.113.10",
+        "203.0.113.10",
+      ]),
+    ).toBeUndefined();
+  });
+
+  it("emits configured mode when the service has a pool", () => {
+    expect(
+      effectiveLbModeForSync("failover", "round_robin", [
+        "203.0.113.10",
+        "203.0.113.20",
+      ]),
+    ).toBe("failover");
+    expect(
+      effectiveLbModeForSync("off", "weighted", [
+        "203.0.113.10",
+        "198.51.100.1",
+      ]),
+    ).toBe("weighted");
+  });
+});
+
+describe("cfdmBindingSyncItemSchema lbMode", () => {
+  const base = {
+    bindingId: 1,
+    serviceId: 10,
+    serviceName: "VPN",
+    serviceSlug: "vpn",
+    fqdn: "vpn.example.com",
+    zoneName: "example.com",
+    hostname: "vpn",
+    ips: ["203.0.113.10"],
+  };
+
+  it("accepts optional lbMode on sync payload", () => {
+    const parsed = cfdmBindingSyncItemSchema.parse({ ...base, lbMode: "failover" });
+    expect(parsed.lbMode).toBe("failover");
+  });
+
+  it("accepts payload without lbMode (legacy)", () => {
+    const parsed = cfdmBindingSyncItemSchema.parse(base);
+    expect(parsed.lbMode).toBeUndefined();
+  });
+});

apps/web/src/components/failover-timeline.tsx

@@ -1,40 +1,161 @@
+import { ShieldCheckIcon } from 'lucide-react'
+
 import {
   Timeline,
   TimelineContent,
+  TimelineDate,
   TimelineHeader,
   TimelineIndicator,
   TimelineItem,
   TimelineSeparator,
   TimelineTitle,
 } from '@/components/reui/timeline'
+import { HealthCheckBadge } from '@/components/health-check-badge'
+import { EmptyState } from '@/components/empty-state'
+import { formatDate, formatRelative, sqliteUtcToIso } from '@/lib/format'
+import {
+  failoverEventCopy,
+  type FailoverEvent,
+  type FailoverHistoryItem,
+} from '@/lib/failover-events'
+import { cn } from '@cfdm/ui/lib/utils'
+import type { ComponentProps } from 'react'
 
-export interface FailoverEvent {
-  id: string
-  title: string
-  detail: string
+type HealthBadgeStatus = ComponentProps<typeof HealthCheckBadge>['status']
+
+function failStreakLabel(count: number): string {
+  const mod10 = count % 10
+  const mod100 = count % 100
+  if (mod10 === 1 && mod100 !== 11) return `${count} ошибка подряд`
+  if (mod10 >= 2 && mod10 <= 4 && (mod100 < 12 || mod100 > 14)) {
+    return `${count} ошибки подряд`
+  }
+  return `${count} ошибок подряд`
 }
 
-export function FailoverTimeline({ events }: { events: FailoverEvent[] }) {
-  if (events.length === 0) {
+function indicatorClass(tone: 'down' | 'added' | 'removed'): string {
+  if (tone === 'added') {
+    return 'border-success bg-success/15 group-data-completed/timeline-item:border-success'
+  }
+  return 'border-destructive bg-destructive/15 group-data-completed/timeline-item:border-destructive'
+}
+
+function separatorClass(tone: 'down' | 'added' | 'removed'): string {
+  if (tone === 'added') return 'bg-success/25'
+  return 'bg-destructive/25'
+}
+
+/**
+ * Failover как sibling «Смены статуса»: ReUI Timeline + Badge.
+ * Preview: https://reui.io/preview/base/components/c-timeline-10
+ * Preview: https://reui.io/preview/base/empty-state-12
+ * Docs: https://reui.io/docs/components/base/timeline
+ * Docs: https://reui.io/docs/components/base/badge
+ */
+export function FailoverTimeline({
+  events,
+  history = [],
+}: {
+  events: FailoverEvent[]
+  history?: readonly FailoverHistoryItem[]
+}) {
+  if (events.length === 0 && history.length === 0) {
     return (
-      <p className="text-muted-foreground text-sm">
-        Событий failover пока нет.
-      </p>
+      <EmptyState
+        icon={ShieldCheckIcon}
+        title="Нет инцидентов"
+        description="Нет Down и нет выходов из общего пула"
+        stackedIcon={false}
+        centered={false}
+      />
     )
   }
 
   return (
-    <Timeline defaultValue={events.length} className="w-full">
-      {events.map((event, index) => (
-        <TimelineItem key={event.id} step={index + 1}>
-          <TimelineSeparator />
-          <TimelineIndicator />
-          <TimelineHeader>
-            <TimelineTitle>{event.title}</TimelineTitle>
-          </TimelineHeader>
-          <TimelineContent>{event.detail}</TimelineContent>
-        </TimelineItem>
-      ))}
-    </Timeline>
+    <div className="flex flex-col gap-4">
+      {events.length > 0 ? (
+        <Timeline defaultValue={0} className="gap-0">
+          {events.map((event, index) => {
+            const checkedIso = event.lastCheckAt
+              ? (sqliteUtcToIso(event.lastCheckAt) ?? event.lastCheckAt)
+              : null
+
+            return (
+              <TimelineItem key={event.id} step={index + 1}>
+                <TimelineSeparator className={separatorClass('down')} />
+                <TimelineIndicator className={indicatorClass('down')} />
+                <TimelineHeader>
+                  <TimelineTitle className="flex flex-wrap items-center gap-2">
+                    <span className="font-mono text-sm">{event.address}</span>
+                    <HealthCheckBadge
+                      status={event.status as HealthBadgeStatus}
+                      lastError={event.lastFailureReason}
+                      lastCheckedAt={checkedIso}
+                      size="xs"
+                    />
+                  </TimelineTitle>
+                  <TimelineDate>
+                    {event.consecutiveFailures > 0
+                      ? failStreakLabel(event.consecutiveFailures)
+                      : null}
+                    {checkedIso
+                      ? `${event.consecutiveFailures > 0 ? ' · ' : ''}${formatRelative(checkedIso)} · ${formatDate(checkedIso)}`
+                      : null}
+                  </TimelineDate>
+                </TimelineHeader>
+                <TimelineContent className="flex flex-col gap-2">
+                  <p className="text-foreground text-sm">
+                    {failoverEventCopy(event)}
+                  </p>
+                  {event.lastFailureReason ? (
+                    <code
+                      className={cn(
+                        'bg-muted block overflow-x-auto rounded-md px-2 py-1.5 font-mono text-xs',
+                      )}
+                    >
+                      {event.lastFailureReason}
+                    </code>
+                  ) : null}
+                </TimelineContent>
+              </TimelineItem>
+            )
+          })}
+        </Timeline>
+      ) : null}
+
+      {history.length > 0 ? (
+        <Timeline defaultValue={0} className="gap-0">
+          {history.map((item, index) => {
+            const checkedIso = sqliteUtcToIso(item.created_at) ?? item.created_at
+            const tone = item.action === 'added' ? 'added' : 'removed'
+
+            return (
+              <TimelineItem key={item.id} step={index + 1}>
+                <TimelineSeparator className={separatorClass(tone)} />
+                <TimelineIndicator className={indicatorClass(tone)} />
+                <TimelineHeader>
+                  <TimelineTitle className="flex flex-wrap items-center gap-2">
+                    <span className="font-mono text-sm">{item.ip}</span>
+                    <HealthCheckBadge
+                      status={item.action === 'added' ? 'up' : 'down'}
+                      size="xs"
+                    />
+                    <span className="text-muted-foreground text-xs">
+                      {item.fqdn}
+                    </span>
+                  </TimelineTitle>
+                  <TimelineDate>
+                    {formatRelative(checkedIso)} · {formatDate(checkedIso)}
+                  </TimelineDate>
+                </TimelineHeader>
+                <TimelineContent>
+                  <p className="text-foreground text-sm">{item.copy}</p>
+                </TimelineContent>
+              </TimelineItem>
+            )
+          })}
+        </Timeline>
+      ) : null}
+    </div>
   )
 }

apps/web/src/components/health-check-badge.tsx

@@ -16,8 +16,8 @@ type HealthStatus =
 
 function normalizeHealth(status: HealthStatus): IpHealthStatus['status'] {
   if (status === 'healthy') return 'up'
-  if (status === 'unhealthy' || status === 'disabled') return 'down'
-  if (status === 'checking') return 'unknown'
+  if (status === 'unhealthy') return 'down'
+  if (status === 'disabled' || status === 'checking') return 'unknown'
   return status
 }
 

apps/web/src/components/health-check-config-fields.tsx

@@ -1,6 +1,7 @@
 import { FormFieldSimple } from '@/components/form-field'
 import { AppInput } from '@/components/app-input'
 import { SettingRow } from '@/components/setting-row'
+import { SelectField } from '@/components/select-field'
 import { Badge } from '@/components/reui/badge'
 import {
   NumberField,
@@ -9,15 +10,8 @@ import {
   NumberFieldIncrement,
   NumberFieldInput,
 } from '@/components/reui/number-field'
-import {
-  Select,
-  SelectContent,
-  SelectItem,
-  SelectTrigger,
-  SelectValue,
-} from '@cfdm/ui/components/select'
 import { Switch } from '@cfdm/ui/components/switch'
-import { FieldGroup } from '@cfdm/ui/components/field'
+import { Field, FieldGroup, FieldLabel } from '@cfdm/ui/components/field'
 import { Button } from '@cfdm/ui/components/button'
 import { ButtonGroup } from '@cfdm/ui/components/button-group'
 import { CableIcon, GlobeIcon } from 'lucide-react'
@@ -59,9 +53,14 @@ export interface LbAndHealthConfig extends HealthCheckConfig {
 const defaultLbModeOptions = [
   { value: 'round_robin', label: 'Round Robin' },
   { value: 'failover', label: 'Failover (приоритет)' },
-  { value: 'weighted', label: 'Weighted (веса)' },
+  { value: 'weighted', label: 'Веса (подмена IP)' },
 ]
 
+export interface LbPoolMetaChange {
+  weight?: number
+  priority?: number
+}
+
 function CompactNumberField({
   id,
   value,
@@ -95,6 +94,89 @@ function CompactNumberField({
   )
 }
 
+function PoolLbMetaFields({
+  idPrefix,
+  mode,
+  ips,
+  weights,
+  priorities,
+  onMetaChange,
+}: {
+  idPrefix: string
+  mode: Exclude<LbMode, 'round_robin'>
+  ips: readonly string[]
+  weights: Record<string, number>
+  priorities: Record<string, number>
+  onMetaChange?: (ip: string, meta: LbPoolMetaChange) => void
+}) {
+  const isWeighted = mode === 'weighted'
+  const minPriority =
+    ips.length === 0
+      ? 1
+      : Math.min(...ips.map((ip) => priorities[ip] ?? 1))
+
+  return (
+    <SettingRow
+      title={isWeighted ? 'Вес IP' : 'Приоритет IP'}
+      description={
+        isWeighted
+          ? 'Доля времени на общем FQDN: 1 и 3 = ¼ и ¾ цикла (слот 60 с)'
+          : '1 — основной, больше — запасной'
+      }
+      compact
+      stacked
+      className="gap-3 px-0 py-3"
+      contentClassName="min-w-0"
+    >
+      {ips.length === 0 ? (
+        <p className="text-muted-foreground text-sm">Сначала добавьте IP выше</p>
+      ) : (
+        <div className="flex w-full flex-col gap-2">
+          {ips.map((ip) => {
+            const isPrimary = (priorities[ip] ?? 1) === minPriority
+            const fieldId = isWeighted
+              ? `${idPrefix}-weight-${ip}`
+              : `${idPrefix}-priority-${ip}`
+            return (
+              <div key={ip} className="flex items-center gap-3">
+                <span className="min-w-0 flex-1 truncate font-mono text-sm">{ip}</span>
+                {!isWeighted ? (
+                  <Badge
+                    variant={isPrimary ? 'success-light' : 'outline'}
+                    size="xs"
+                    radius="full"
+                  >
+                    {isPrimary ? 'Основной' : 'Запасной'}
+                  </Badge>
+                ) : null}
+                <Field className="w-28 gap-0">
+                  <FieldLabel htmlFor={fieldId} className="sr-only">
+                    {isWeighted ? `Вес ${ip}` : `Приоритет ${ip}`}
+                  </FieldLabel>
+                  <CompactNumberField
+                    id={fieldId}
+                    value={isWeighted ? (weights[ip] ?? 1) : (priorities[ip] ?? 1)}
+                    min={1}
+                    max={100}
+                    onValueChange={(next) =>
+                      onMetaChange?.(
+                        ip,
+                        isWeighted
+                          ? { weight: next ?? 1 }
+                          : { priority: next ?? 1 },
+                      )
+                    }
+                  />
+                </Field>
+              </div>
+            )
+          })}
+        </div>
+      )}
+    </SettingRow>
+  )
+}
+
 export function HealthCheckConfigFields({
   value,
   onChange,
@@ -102,6 +184,10 @@ export function HealthCheckConfigFields({
   lbModeOptions = defaultLbModeOptions,
   idPrefix = 'health',
   showLbMode = true,
+  ips = [],
+  weights = {},
+  priorities = {},
+  onMetaChange,
   className,
 }: {
   value: LbAndHealthConfig
@@ -110,6 +196,10 @@ export function HealthCheckConfigFields({
   lbModeOptions?: { value: string; label: string }[]
   idPrefix?: string
   showLbMode?: boolean
+  ips?: readonly string[]
+  weights?: Record<string, number>
+  priorities?: Record<string, number>
+  onMetaChange?: (ip: string, meta: LbPoolMetaChange) => void
   className?: string
 }) {
   function patch(next: Partial<LbAndHealthConfig>) {
@@ -134,25 +224,28 @@ export function HealthCheckConfigFields({
           compact
           className={rowClass}
         >
-          <Select
+          <SelectField
             modal={false}
             value={value.lb_mode}
             onValueChange={(v) => patch({ lb_mode: (v ?? 'round_robin') as LbMode })}
-          >
-            <SelectTrigger id={`${idPrefix}-lb-mode`} className="w-full">
-              <SelectValue placeholder="Выберите режим" />
-            </SelectTrigger>
-            <SelectContent>
-              {lbModeOptions.map((item) => (
-                <SelectItem key={item.value} value={item.value}>
-                  {item.label}
-                </SelectItem>
-              ))}
-            </SelectContent>
-          </Select>
+            triggerId={`${idPrefix}-lb-mode`}
+            placeholder="Выберите режим"
+            options={lbModeOptions}
+          />
         </SettingRow>
       ) : null}
 
+      {showLbMode && value.lb_mode !== 'round_robin' ? (
+        <PoolLbMetaFields
+          idPrefix={idPrefix}
+          mode={value.lb_mode}
+          ips={ips}
+          weights={weights}
+          priorities={priorities}
+          onMetaChange={onMetaChange}
+        />
+      ) : null}
+
       <SettingRow
         title="Провайдер health-check"
         description="Кто пробирует цель. Можно выбрать несколько источников."

apps/web/src/components/reui-kit/index.ts

@@ -1,5 +1,6 @@
 export { UptimeChart, type UptimeProbe, type UptimePeriodKey, probeUptimePercent, lastProbeLatency } from './uptime-chart'
 export { ServiceHealthMonitor } from './service-health-monitor'
+export { ServiceFailoverPanel } from './service-failover-panel'
 export { applyFiltersToData, getActiveFilters, renderSingleSelectedLabel } from './filter-utils'
 export { CertStatusChart, GroupDomainsChart } from './dashboard-analytics'
 export { ResourcePage, type ResourcePageProps, type ResourcePageTab } from './resource-page'

apps/web/src/components/reui-kit/service-address-block.tsx

@@ -1,8 +1,7 @@
-import { useState, type KeyboardEvent } from 'react'
-import { PlusIcon, ServerIcon, Trash2Icon } from 'lucide-react'
+import { useState, type KeyboardEvent, type ReactNode } from 'react'
+import { ServerIcon, Trash2Icon } from 'lucide-react'
 
 import { EmptyState } from '@/components/empty-state'
-import { ServiceBindingIpInput } from '@/components/service-binding-ip-input'
 import { isValidIpv4 } from '@/components/tagged-input'
 import { Badge } from '@/components/reui/badge'
 import {
@@ -16,15 +15,18 @@ import { IconTile } from '@/components/reui/icon-tile'
 import { parseFqdn } from '@/lib/parse-fqdn'
 import {
   addAddressNode,
-  emptyBindingDraft,
+  addCommonFqdn,
+  addExtraFqdn,
+  addressHasFqdn,
   removeAddressNode,
-  withPoolIps,
+  removeCommonFqdn,
+  removeExtraFqdn,
+  updateCommonFqdn,
+  updateExtraFqdn,
   type AddressBlockState,
-  type ServiceBindingDraft,
 } from '@/lib/service-address'
 import { Button } from '@cfdm/ui/components/button'
 import { Field, FieldLabel } from '@cfdm/ui/components/field'
-import { Input } from '@cfdm/ui/components/input'
 import {
   InputGroup,
   InputGroupAddon,
@@ -39,16 +41,36 @@ import {
   ItemMedia,
   ItemTitle,
 } from '@cfdm/ui/components/item'
-import {
-  Select,
-  SelectContent,
-  SelectItem,
-  SelectTrigger,
-  SelectValue,
-} from '@cfdm/ui/components/select'
+
+function ZoneAddon({
+  fqdn,
+  zoneHints,
+  trailing,
+}: {
+  fqdn: string
+  zoneHints: string[]
+  trailing?: ReactNode
+}) {
+  const parsed = parseFqdn(fqdn, zoneHints)
+  if (!parsed && !fqdn.trim() && !trailing) return null
+  return (
+    <InputGroupAddon align="inline-end">
+      {parsed ? (
+        <Badge variant="outline" size="xs" className="font-mono">
+          {parsed.zoneName}
+        </Badge>
+      ) : fqdn.trim() ? (
+        <Badge variant="warning-light" size="xs">
+          зона не найдена
+        </Badge>
+      ) : null}
+      {trailing}
+    </InputGroupAddon>
+  )
+}
 
 /**
- * Единый блок адресов сервиса: общий FQDN + пул IP с опциональным доп. доменом.
+ * Единый блок адресов сервиса: список общих FQDN на весь пул + IP с доп. доменами.
  * Preview: https://reui.io/preview/base/settings-3
  * Preview: https://reui.io/preview/base/list-9
  * Preview: https://reui.io/preview/base/form-7
@@ -67,17 +89,32 @@ export function ServiceAddressBlock({
 }) {
   const [pendingIp, setPendingIp] = useState('')
   const [ipInvalid, setIpInvalid] = useState(false)
-  const [otherOpen, setOtherOpen] = useState(value.otherBindings.length > 0)
+  const [pendingFqdn, setPendingFqdn] = useState('')
+  const [fqdnInvalid, setFqdnInvalid] = useState(false)
+  const [pendingExtraByIp, setPendingExtraByIp] = useState<Record<string, string>>({})
+  const [extraInvalidByIp, setExtraInvalidByIp] = useState<Record<string, boolean>>({})
 
   const pool = value.nodes.map((node) => node.ip)
-  const parsedCommon = parseFqdn(value.commonFqdn, zoneHints)
-  const showOthers = otherOpen || value.otherBindings.length > 0
-  const pendingTrimmed = pendingIp.trim()
-  const pendingInvalid =
-    ipInvalid && pendingTrimmed.length > 0 && !isValidIpv4(pendingTrimmed)
+  const pendingIpTrimmed = pendingIp.trim()
+  const pendingFqdnTrimmed = pendingFqdn.trim()
+  const pendingIpInvalid =
+    ipInvalid && pendingIpTrimmed.length > 0 && !isValidIpv4(pendingIpTrimmed)
+  const pendingFqdnInvalid =
+    fqdnInvalid && pendingFqdnTrimmed.length > 0
 
-  function handleCommonFqdn(next: string) {
-    onChange({ ...value, commonFqdn: next })
+  function tryAddFqdn(raw: string) {
+    const trimmed = raw.trim()
+    if (!trimmed) {
+      setFqdnInvalid(false)
+      return
+    }
+    if (addressHasFqdn(value, trimmed)) {
+      setFqdnInvalid(true)
+      return
+    }
+    onChange(addCommonFqdn(value, trimmed))
+    setPendingFqdn('')
+    setFqdnInvalid(false)
   }
 
   function tryAddIp(raw: string) {
@@ -95,45 +132,40 @@ export function ServiceAddressBlock({
     setIpInvalid(false)
   }
 
-  function handlePendingKeyDown(event: KeyboardEvent<HTMLInputElement>) {
+  function handleFqdnKeyDown(event: KeyboardEvent<HTMLInputElement>) {
+    if (event.key === 'Enter') {
+      event.preventDefault()
+      tryAddFqdn(pendingFqdn)
+    }
+  }
+
+  function handleIpKeyDown(event: KeyboardEvent<HTMLInputElement>) {
     if (event.key === 'Enter') {
       event.preventDefault()
       tryAddIp(pendingIp)
     }
   }
 
-  function handleNodeFqdn(ip: string, extraFqdn: string) {
-    onChange({
-      ...value,
-      nodes: value.nodes.map((node) =>
-        node.ip === ip ? { ...node, extraFqdn } : node,
-      ),
-    })
+  function tryAddExtra(ip: string, raw: string) {
+    const trimmed = raw.trim()
+    if (!trimmed) {
+      setExtraInvalidByIp((current) => ({ ...current, [ip]: false }))
+      return
+    }
+    if (addressHasFqdn(value, trimmed)) {
+      setExtraInvalidByIp((current) => ({ ...current, [ip]: true }))
+      return
+    }
+    onChange(addExtraFqdn(value, ip, trimmed))
+    setPendingExtraByIp((current) => ({ ...current, [ip]: '' }))
+    setExtraInvalidByIp((current) => ({ ...current, [ip]: false }))
   }
 
-  function handleRemoveIp(ip: string) {
-    onChange(removeAddressNode(value, ip))
-  }
-
-  function handleAddOther() {
-    setOtherOpen(true)
-    onChange({
-      ...value,
-      otherBindings: [...value.otherBindings, withPoolIps(emptyBindingDraft(), pool)],
-    })
-  }
-
-  function handleOtherChange(index: number, next: ServiceBindingDraft) {
-    onChange({
-      ...value,
-      otherBindings: value.otherBindings.map((item, i) => (i === index ? next : item)),
-    })
-  }
-
-  function handleRemoveOther(index: number) {
-    const otherBindings = value.otherBindings.filter((_, i) => i !== index)
-    onChange({ ...value, otherBindings })
-    if (otherBindings.length === 0) setOtherOpen(false)
+  function handleExtraKeyDown(ip: string, event: KeyboardEvent<HTMLInputElement>) {
+    if (event.key === 'Enter') {
+      event.preventDefault()
+      tryAddExtra(ip, pendingExtraByIp[ip] ?? '')
+    }
   }
 
   return (
@@ -142,33 +174,63 @@ export function ServiceAddressBlock({
         <FrameHeader className="px-0 pt-0">
           <FrameTitle>Адреса</FrameTitle>
           <FrameDescription>
-            Общий FQDN на весь пул · у каждого IP свой доп. домен
+            Общие FQDN — на весь пул · у IP свои доп. домены
           </FrameDescription>
         </FrameHeader>
         <Field>
-          <FieldLabel htmlFor="service-common-fqdn">Общий домен (FQDN)</FieldLabel>
-          <InputGroup>
-            <InputGroupInput
-              id="service-common-fqdn"
-              className="font-mono"
-              value={value.commonFqdn}
-              placeholder={zoneHints[0] ? `gw.${zoneHints[0]}` : 'gw.ivx.su'}
-              onChange={(event) => handleCommonFqdn(event.target.value)}
-            />
-            {parsedCommon ? (
-              <InputGroupAddon align="inline-end">
-                <Badge variant="outline" size="xs" className="font-mono">
-                  {parsedCommon.zoneName}
-                </Badge>
-              </InputGroupAddon>
-            ) : value.commonFqdn.trim() ? (
-              <InputGroupAddon align="inline-end">
-                <Badge variant="warning-light" size="xs">
-                  зона не найдена
-                </Badge>
-              </InputGroupAddon>
-            ) : null}
-          </InputGroup>
+          <FieldLabel htmlFor="service-common-fqdn-add">Общие домены (FQDN)</FieldLabel>
+          <div className="flex w-full flex-col gap-2">
+            {value.commonFqdns.map((fqdn, index) => (
+              <InputGroup key={`common-fqdn-${index}`}>
+                <InputGroupInput
+                  id={`service-common-fqdn-${index}`}
+                  className="font-mono"
+                  value={fqdn}
+                  placeholder={zoneHints[0] ? `gw.${zoneHints[0]}` : 'gw.ivx.su'}
+                  onChange={(event) =>
+                    onChange(updateCommonFqdn(value, index, event.target.value))
+                  }
+                />
+                <ZoneAddon
+                  fqdn={fqdn}
+                  zoneHints={zoneHints}
+                  trailing={
+                    <InputGroupButton
+                      size="icon-xs"
+                      aria-label={`Удалить ${fqdn || 'FQDN'}`}
+                      onClick={() => onChange(removeCommonFqdn(value, index))}
+                    >
+                      <Trash2Icon />
+                    </InputGroupButton>
+                  }
+                />
+              </InputGroup>
+            ))}
+            <InputGroup>
+              <InputGroupInput
+                id="service-common-fqdn-add"
+                className="font-mono"
+                value={pendingFqdn}
+                placeholder={zoneHints[0] ? `gw.${zoneHints[0]}` : 'gw.ivx.su'}
+                aria-invalid={pendingFqdnInvalid || undefined}
+                onChange={(event) => {
+                  setPendingFqdn(event.target.value)
+                  setFqdnInvalid(false)
+                }}
+                onKeyDown={handleFqdnKeyDown}
+                onBlur={() => tryAddFqdn(pendingFqdn)}
+              />
+              <ZoneAddon
+                fqdn={pendingFqdn}
+                zoneHints={zoneHints}
+                trailing={
+                  <InputGroupButton size="sm" onClick={() => tryAddFqdn(pendingFqdn)}>
+                    Добавить
+                  </InputGroupButton>
+                }
+              />
+            </InputGroup>
+          </div>
         </Field>
       </FramePanel>
 
@@ -180,14 +242,13 @@ export function ServiceAddressBlock({
           <EmptyState
             icon={ServerIcon}
             title="Добавьте IP пула"
-            description="IPv4 сервиса. Для каждого адреса можно указать доп. FQDN."
+            description="IPv4 сервиса. Для каждого адреса можно указать несколько доп. FQDN, в том числе wildcard."
             stackedIcon={false}
             centered={false}
           />
         ) : (
           <ItemGroup className="gap-2">
             {value.nodes.map((node) => {
-              const parsedExtra = parseFqdn(node.extraFqdn, zoneHints)
               return (
                 <Item
                   key={node.ip}
@@ -214,7 +275,7 @@ export function ServiceAddressBlock({
                           variant="ghost"
                           size="icon-sm"
                           aria-label={`Удалить ${node.ip}`}
-                          onClick={() => handleRemoveIp(node.ip)}
+                          onClick={() => onChange(removeAddressNode(value, node.ip))}
                         >
                           <Trash2Icon />
                         </Button>
@@ -222,39 +283,98 @@ export function ServiceAddressBlock({
                     </div>
                     <Field className="gap-1.5">
                       <FieldLabel
-                        htmlFor={`service-ip-extra-${node.ip}`}
+                        htmlFor={`service-ip-extra-add-${node.ip}`}
                         className="text-muted-foreground text-xs"
                       >
                         Доп. FQDN
                       </FieldLabel>
-                      <InputGroup>
-                        <InputGroupInput
-                          id={`service-ip-extra-${node.ip}`}
-                          className="font-mono"
-                          value={node.extraFqdn}
-                          placeholder={
-                            zoneHints[0]
-                              ? `необязательно · spb.${zoneHints[0]}`
-                              : 'необязательно · spb.example.com'
-                          }
-                          onChange={(event) =>
-                            handleNodeFqdn(node.ip, event.target.value)
-                          }
-                        />
-                        {parsedExtra ? (
-                          <InputGroupAddon align="inline-end">
-                            <Badge variant="outline" size="xs" className="font-mono">
-                              {parsedExtra.zoneName}
-                            </Badge>
-                          </InputGroupAddon>
-                        ) : node.extraFqdn.trim() ? (
-                          <InputGroupAddon align="inline-end">
-                            <Badge variant="warning-light" size="xs">
-                              зона не найдена
-                            </Badge>
-                          </InputGroupAddon>
-                        ) : null}
-                      </InputGroup>
+                      <div className="flex w-full flex-col gap-2">
+                        {node.extraFqdns.map((fqdn, index) => (
+                          <InputGroup key={`extra-fqdn-${node.ip}-${index}`}>
+                            <InputGroupInput
+                              id={`service-ip-extra-${node.ip}-${index}`}
+                              className="font-mono"
+                              value={fqdn}
+                              placeholder={
+                                zoneHints[0]
+                                  ? `*.mdns.${zoneHints[0]}`
+                                  : '*.mdns.example.com'
+                              }
+                              onChange={(event) =>
+                                onChange(
+                                  updateExtraFqdn(
+                                    value,
+                                    node.ip,
+                                    index,
+                                    event.target.value,
+                                  ),
+                                )
+                              }
+                            />
+                            <ZoneAddon
+                              fqdn={fqdn}
+                              zoneHints={zoneHints}
+                              trailing={
+                                <InputGroupButton
+                                  size="icon-xs"
+                                  aria-label={`Удалить ${fqdn || 'FQDN'}`}
+                                  onClick={() =>
+                                    onChange(removeExtraFqdn(value, node.ip, index))
+                                  }
+                                >
+                                  <Trash2Icon />
+                                </InputGroupButton>
+                              }
+                            />
+                          </InputGroup>
+                        ))}
+                        <InputGroup>
+                          <InputGroupInput
+                            id={`service-ip-extra-add-${node.ip}`}
+                            className="font-mono"
+                            value={pendingExtraByIp[node.ip] ?? ''}
+                            placeholder={
+                              zoneHints[0]
+                                ? `необязательно · *.mdns.${zoneHints[0]}`
+                                : 'необязательно · *.mdns.example.com'
+                            }
+                            aria-invalid={
+                              extraInvalidByIp[node.ip] &&
+                              (pendingExtraByIp[node.ip] ?? '').trim().length > 0
+                                ? true
+                                : undefined
+                            }
+                            onChange={(event) => {
+                              setPendingExtraByIp((current) => ({
+                                ...current,
+                                [node.ip]: event.target.value,
+                              }))
+                              setExtraInvalidByIp((current) => ({
+                                ...current,
+                                [node.ip]: false,
+                              }))
+                            }}
+                            onKeyDown={(event) => handleExtraKeyDown(node.ip, event)}
+                            onBlur={() =>
+                              tryAddExtra(node.ip, pendingExtraByIp[node.ip] ?? '')
+                            }
+                          />
+                          <ZoneAddon
+                            fqdn={pendingExtraByIp[node.ip] ?? ''}
+                            zoneHints={zoneHints}
+                            trailing={
+                              <InputGroupButton
+                                size="sm"
+                                onClick={() =>
+                                  tryAddExtra(node.ip, pendingExtraByIp[node.ip] ?? '')
+                                }
+                              >
+                                Добавить
+                              </InputGroupButton>
+                            }
+                          />
+                        </InputGroup>
+                      </div>
                     </Field>
                   </ItemContent>
                 </Item>
@@ -268,12 +388,12 @@ export function ServiceAddressBlock({
             className="font-mono"
             value={pendingIp}
             placeholder="192.168.1.1"
-            aria-invalid={pendingInvalid || undefined}
+            aria-invalid={pendingIpInvalid || undefined}
             onChange={(event) => {
               setPendingIp(event.target.value)
               setIpInvalid(false)
             }}
-            onKeyDown={handlePendingKeyDown}
+            onKeyDown={handleIpKeyDown}
             onBlur={() => tryAddIp(pendingIp)}
           />
           <InputGroupAddon align="inline-end">
@@ -282,156 +402,7 @@ export function ServiceAddressBlock({
             </InputGroupButton>
           </InputGroupAddon>
         </InputGroup>
-        {showOthers ? null : (
-          <div className="flex justify-end">
-            <Button
-              type="button"
-              variant="ghost"
-              size="sm"
-              onClick={handleAddOther}
-            >
-              <PlusIcon data-icon="inline-start" />
-              Другой FQDN
-            </Button>
-          </div>
-        )}
       </FramePanel>
-
-      {showOthers ? (
-        <FramePanel fit className="flex flex-col gap-3">
-          <FrameHeader className="flex flex-row items-start justify-between gap-2 px-0 pt-0">
-            <div className="flex min-w-0 flex-col gap-1">
-              <FrameTitle>Другие FQDN</FrameTitle>
-              <FrameDescription>CNAME и A не 1:1 с IP пула</FrameDescription>
-            </div>
-            <Button type="button" variant="outline" size="sm" onClick={handleAddOther}>
-              <PlusIcon data-icon="inline-start" />
-              Добавить
-            </Button>
-          </FrameHeader>
-          {value.otherBindings.length === 0 ? (
-            <p className="text-muted-foreground text-sm">Нет дополнительных FQDN</p>
-          ) : (
-            <ItemGroup className="gap-2">
-              {value.otherBindings.map((binding, index) => {
-                const parsedZone = parseFqdn(binding.fqdn, zoneHints)
-                return (
-                  <Item
-                    key={`other-binding-${index}`}
-                    variant="outline"
-                    size="sm"
-                    className="items-stretch"
-                  >
-                    <ItemContent className="flex w-full min-w-0 flex-col gap-2">
-                      <div className="flex items-center gap-2">
-                        {parsedZone ? (
-                          <Badge variant="outline" size="xs" className="font-mono">
-                            {parsedZone.zoneName}
-                          </Badge>
-                        ) : binding.fqdn.trim() ? (
-                          <Badge variant="warning-light" size="xs">
-                            зона не найдена
-                          </Badge>
-                        ) : (
-                          <span className="text-muted-foreground text-xs">FQDN</span>
-                        )}
-                        <Button
-                          type="button"
-                          variant="ghost"
-                          size="icon-sm"
-                          className="ml-auto shrink-0"
-                          aria-label="Удалить FQDN"
-                          onClick={() => handleRemoveOther(index)}
-                        >
-                          <Trash2Icon />
-                        </Button>
-                      </div>
-                      <div className="grid grid-cols-1 gap-2 sm:grid-cols-[minmax(0,1fr)_7.5rem]">
-                        <Input
-                          id={`other-fqdn-${index}`}
-                          className="font-mono"
-                          value={binding.fqdn}
-                          onChange={(event) =>
-                            handleOtherChange(index, {
-                              ...binding,
-                              fqdn: event.target.value,
-                            })
-                          }
-                          placeholder={
-                            zoneHints[0] ? `api.${zoneHints[0]}` : 'api.ivx.su'
-                          }
-                        />
-                        <Select
-                          items={[
-                            { label: 'A (IP)', value: 'A' },
-                            { label: 'CNAME', value: 'CNAME' },
-                          ]}
-                          value={binding.record_type}
-                          onValueChange={(next) => {
-                            const recordType = (next ?? 'A') as 'A' | 'CNAME'
-                            handleOtherChange(index, {
-                              ...binding,
-                              record_type: recordType,
-                              target_ips: recordType === 'A' ? binding.target_ips : [],
-                              target_cname:
-                                recordType === 'CNAME' ? binding.target_cname : '',
-                            })
-                          }}
-                        >
-                          <SelectTrigger id={`other-type-${index}`} className="w-full">
-                            <SelectValue />
-                          </SelectTrigger>
-                          <SelectContent>
-                            <SelectItem value="A">A (IP)</SelectItem>
-                            <SelectItem value="CNAME">CNAME</SelectItem>
-                          </SelectContent>
-                        </Select>
-                      </div>
-                      {binding.record_type === 'CNAME' ? (
-                        <Input
-                          id={`other-cname-${index}`}
-                          value={binding.target_cname}
-                          placeholder="mmsk.rkns.top"
-                          onChange={(event) =>
-                            handleOtherChange(index, {
-                              ...binding,
-                              target_cname: event.target.value,
-                            })
-                          }
-                        />
-                      ) : (
-                        <ServiceBindingIpInput
-                          id={`other-ip-${index}`}
-                          value={binding.target_ips}
-                          pool={pool}
-                          onChange={(targetIps) =>
-                            handleOtherChange(index, {
-                              ...binding,
-                              target_ips: targetIps,
-                              target_ip_weights: Object.fromEntries(
-                                targetIps.map((ip) => [
-                                  ip,
-                                  binding.target_ip_weights[ip] ?? 1,
-                                ]),
-                              ),
-                              target_ip_priorities: Object.fromEntries(
-                                targetIps.map((ip) => [
-                                  ip,
-                                  binding.target_ip_priorities[ip] ?? 1,
-                                ]),
-                              ),
-                            })
-                          }
-                        />
-                      )}
-                    </ItemContent>
-                  </Item>
-                )
-              })}
-            </ItemGroup>
-          )}
-        </FramePanel>
-      ) : null}
     </Frame>
   )
 }

apps/web/src/components/reui-kit/service-failover-panel.tsx

@@ -0,0 +1,148 @@
+import { UnplugIcon } from 'lucide-react'
+
+import { FailoverTimeline } from '@/components/failover-timeline'
+import {
+  mergeFailoverHistory,
+  toFailoverEvents,
+  type FailoverBindingPool,
+  type FailoverHealthInput,
+} from '@/lib/failover-events'
+import {
+  latestHealthByIp,
+  resolveIpDisplayHealth,
+  type HealthLogProbe,
+  type HealthLogStatus,
+} from '@/lib/health-log'
+import type { FailoverLogEntry, ServiceView } from '@/lib/schemas'
+import { Badge } from '@/components/reui/badge'
+import {
+  Frame,
+  FrameDescription,
+  FrameHeader,
+  FramePanel,
+  FrameTitle,
+} from '@/components/reui/frame'
+import {
+  Alert,
+  AlertDescription,
+  AlertTitle,
+} from '@/components/reui/alert'
+
+type LbMode = ServiceView['lb_mode']
+
+const PANEL_COPY: Record<LbMode, { title: string; description: string }> = {
+  failover: {
+    title: 'Failover (приоритет)',
+    description: 'Down снимается с общего FQDN. История: кто вышел из пула и кто вернулся',
+  },
+  weighted: {
+    title: 'Веса (подмена IP)',
+    description: 'На общем FQDN один IP по весам. Down выводится из цикла',
+  },
+  round_robin: {
+    title: 'Round Robin',
+    description: 'На общем FQDN все живые A. Down снимается с пула',
+  },
+}
+
+function failoverCountLabel(count: number): string {
+  const mod10 = count % 10
+  const mod100 = count % 100
+  if (mod10 === 1 && mod100 !== 11) return `${count} адрес Down`
+  if (mod10 >= 2 && mod10 <= 4 && (mod100 < 12 || mod100 > 14)) {
+    return `${count} адреса Down`
+  }
+  return `${count} адресов Down`
+}
+
+function alertDescription(events: { kind: string; fqdns: string[] }[]): string {
+  const removedCount = events.filter((event) => event.kind === 'removed').length
+  if (removedCount > 0) return 'Сняты с общего FQDN'
+  if (events.some((event) => event.fqdns.length > 0)) {
+    return 'Остались в A-записях общего FQDN как last-resort'
+  }
+  return 'Down: персональные FQDN не меняются'
+}
+
+/**
+ * Балансировка — текущие Down + кто вышел из общего пула.
+ * Preview: https://reui.io/preview/base/components/c-timeline-10
+ * Preview: https://reui.io/preview/base/empty-state-12
+ * Docs: https://reui.io/docs/components/base/frame
+ * Docs: https://reui.io/docs/components/base/timeline
+ * Docs: https://reui.io/docs/components/base/badge
+ * Docs: https://reui.io/docs/components/base/alert
+ */
+export function ServiceFailoverPanel({
+  lbMode = 'round_robin',
+  hasPool = true,
+  ipHealth,
+  bindings,
+  history,
+  probes = [],
+}: {
+  lbMode?: LbMode
+  hasPool?: boolean
+  ipHealth: readonly FailoverHealthInput[]
+  bindings: readonly FailoverBindingPool[]
+  history: readonly FailoverLogEntry[]
+  probes?: readonly HealthLogProbe[]
+}) {
+  const copy = hasPool
+    ? (PANEL_COPY[lbMode] ?? PANEL_COPY.round_robin)
+    : {
+        title: 'без резервирования',
+        description: 'Один origin IP — балансировка не применяется',
+      }
+  const liveByIp = latestHealthByIp(probes)
+  const overlayHealth = ipHealth.map((row) => {
+    const live = liveByIp.get(row.ip)
+    return {
+      ...row,
+      status: resolveIpDisplayHealth(
+        row.status as HealthLogStatus,
+        live?.status,
+      ),
+      last_error:
+        live && live.status !== 'unknown' ? live.last_error : row.last_error,
+      last_checked_at:
+        live && live.status !== 'unknown'
+          ? live.last_checked_at
+          : row.last_checked_at,
+    }
+  })
+  const events = toFailoverEvents(overlayHealth, bindings)
+  const mergedHistory = mergeFailoverHistory(history, probes, bindings)
+
+  return (
+    <Frame stacked spacing="sm" className="min-w-0 w-full">
+      <FramePanel className="flex flex-col gap-3">
+        <FrameHeader className="gap-1 px-0 py-0">
+          <FrameTitle className="flex flex-wrap items-center gap-2">
+            {copy.title}
+            {events.length > 0 ? (
+              <Badge variant="destructive-light" size="xs" radius="full">
+                {events.length}
+              </Badge>
+            ) : (
+              <Badge variant="success-light" size="xs" radius="full">
+                OK
+              </Badge>
+            )}
+          </FrameTitle>
+          <FrameDescription>{copy.description}</FrameDescription>
+        </FrameHeader>
+
+        {events.length > 0 ? (
+          <Alert variant="destructive">
+            <UnplugIcon aria-hidden="true" />
+            <AlertTitle>{failoverCountLabel(events.length)}</AlertTitle>
+            <AlertDescription>{alertDescription(events)}</AlertDescription>
+          </Alert>
+        ) : null}
+
+        <FailoverTimeline events={events} history={mergedHistory} />
+      </FramePanel>
+    </Frame>
+  )
+}

apps/web/src/components/reui-kit/uptime-chart.test.ts

@@ -45,6 +45,28 @@ describe('toAlignedSeries', () => {
     expect(keys).toEqual(['local', 'cloudflare', 'globalping'])
   })
 
+  it('keeps live latency when another IP in the same bucket is down', () => {
+    const { points } = toAlignedSeries([
+      probe({
+        id: 1,
+        latency_ms: 18,
+        checked_at: '2026-01-01T00:00:10.000Z',
+      }),
+      probe({
+        id: 2,
+        status: 'down',
+        ok: false,
+        latency_ms: null,
+        checked_at: '2026-01-01T00:00:12.000Z',
+      }),
+    ])
+
+    expect(points).toHaveLength(1)
+    expect(points[0]?.local).toBe(18)
+    expect(points[0]?.localOk).toBe(true)
+    expect(points[0]?.ok).toBe(true)
+  })
+
   it('does not plot down probes as latency 0', () => {
     const { points } = toAlignedSeries([
       probe({

apps/web/src/components/reui-kit/uptime-chart.tsx

@@ -124,8 +124,13 @@ export function toAlignedSeries(items: UptimeProbe[]): {
     }
 
     const ok = probeOk(item)
-    row[`${key}Ok`] = ok
-    row[key] = ok ? item.latency_ms : null
+    const prevOk = row[`${key}Ok`]
+    row[`${key}Ok`] = prevOk === true || ok
+    if (ok && item.latency_ms != null) {
+      row[key] = item.latency_ms
+    } else if (row[key] === undefined) {
+      row[key] = null
+    }
   }
 
   const points = [...buckets.entries()]

apps/web/src/components/service-edit-sheet.tsx

@@ -17,6 +17,7 @@ import {
   DEFAULT_BINDING_HEALTH,
   emptyAddressBlock,
   hydrateAddressBlock,
+  patchAddressIpMeta,
   toBindingDrafts,
   toDomainsPayload,
   type AddressBlockState,
@@ -218,8 +219,8 @@ export function ServiceEditSheet({
         <SheetHeader className="shrink-0 border-b pb-4">
           <SheetTitle>{isCreate ? 'Новый сервис' : 'Редактирование сервиса'}</SheetTitle>
           <SheetDescription>
-            Общий домен и пул IP — в одном блоке. У каждого адреса можно указать
-            свой доп. FQDN.
+            Общие FQDN на весь пул IP. У каждого адреса можно указать несколько доп.
+            FQDN, в том числе wildcard.
           </SheetDescription>
         </SheetHeader>
 
@@ -282,6 +283,12 @@ export function ServiceEditSheet({
               idPrefix="service-health"
               value={primaryHealthValue}
               onChange={handlePrimaryHealthChange}
+              ips={address.nodes.map((node) => node.ip)}
+              weights={address.target_ip_weights}
+              priorities={address.target_ip_priorities}
+              onMetaChange={(ip, meta) =>
+                setAddress((current) => patchAddressIpMeta(current, ip, meta))
+              }
             />
           </section>
         </div>

apps/web/src/components/services/service-detail-grid.tsx

@@ -18,6 +18,11 @@ import { DataGridColumnHeader } from '@/components/reui/data-grid/data-grid-colu
 import { IconTile } from '@/components/reui/icon-tile'
 import { createFilter, type Filter, type FilterFieldConfig } from '@/components/reui/filters'
 import { ResourcePage } from '@/components/reui-kit'
+import {
+  latestHealthByIp,
+  resolveIpDisplayHealth,
+  type HealthLogProbe,
+} from '@/lib/health-log'
 import { certRelativeBadge } from '@/components/columns/certificates-columns'
 import { certMonitoringOptions } from '@/lib/cert-monitoring'
 import { formatDate } from '@/lib/format'
@@ -99,8 +104,12 @@ function mapNodeHealth(status: string): HealthStatus {
   return 'unknown'
 }
 
-function buildIpRows(service: ServiceView): ServiceIpRow[] {
+function buildIpRows(
+  service: ServiceView,
+  probes: readonly HealthLogProbe[] = [],
+): ServiceIpRow[] {
   const healthByIp = new Map(service.ip_health.map((row) => [row.ip, row]))
+  const liveByIp = latestHealthByIp(probes)
   const weights = Object.assign(
     {},
     ...service.domains.map((domain) => domain.target_ip_weights ?? {}),
@@ -113,19 +122,25 @@ function buildIpRows(service: ServiceView): ServiceIpRow[] {
 
   return service.ips.map((ip) => {
     const health = healthByIp.get(ip)
+    const live = liveByIp.get(ip)
+    const status = resolveIpDisplayHealth(health?.status, live?.status)
+    const extras = live && live.status !== 'unknown' ? live : health
     return {
       id: ip,
       ip,
-      status: health?.status ?? 'unknown',
+      status,
       enabled: service.ip_enabled[ip] !== false,
       active: activeSet.has(ip),
       weight: weights[ip] ?? 1,
       priority: priorities[ip] ?? 1,
-      latency_ms: health?.latency_ms ?? null,
-      last_checked_at: health?.last_checked_at ?? null,
-      last_error: health?.last_error ?? null,
-      colo: health?.colo ?? null,
-      provider: health?.provider ?? null,
+      latency_ms: extras?.latency_ms ?? null,
+      last_checked_at: extras?.last_checked_at ?? null,
+      last_error:
+        live && live.status !== 'unknown'
+          ? live.last_error
+          : (health?.last_error ?? null),
+      colo: extras?.colo ?? null,
+      provider: extras?.provider ?? null,
     }
   })
 }
@@ -151,17 +166,23 @@ function buildNodeRows(
     weight: number
     priority: number
   }>,
+  probes: readonly HealthLogProbe[] = [],
 ): ServiceNodeRow[] {
-  return nodes.map((node) => ({
-    id: String(node.id),
-    nodeId: node.id,
-    address: node.address,
-    protocol: node.protocol,
-    port: node.port,
-    health_status: mapNodeHealth(node.health_status),
-    weight: node.weight,
-    priority: node.priority,
-  }))
+  const liveByIp = latestHealthByIp(probes)
+  return nodes.map((node) => {
+    const stored = mapNodeHealth(node.health_status)
+    const live = liveByIp.get(node.address)
+    return {
+      id: String(node.id),
+      nodeId: node.id,
+      address: node.address,
+      protocol: node.protocol,
+      port: node.port,
+      health_status: resolveIpDisplayHealth(stored, live?.status),
+      weight: node.weight,
+      priority: node.priority,
+    }
+  })
 }
 
 function NameCell({
@@ -199,6 +220,7 @@ interface ServiceDetailGridProps {
     weight: number
     priority: number
   }>
+  probes?: readonly HealthLogProbe[]
   togglingIp: string | null
   onToggleIp: (ip: string, enabled: boolean) => void
   onChangeIp: (row: ServiceFqdnRow) => void
@@ -211,6 +233,7 @@ interface ServiceDetailGridProps {
 export function ServiceDetailGrid({
   service,
   nodes,
+  probes = [],
   togglingIp,
   onToggleIp,
   onChangeIp,
@@ -283,9 +306,9 @@ export function ServiceDetailGrid({
     },
   })
 
-  const ipRows = useMemo(() => buildIpRows(service), [service])
+  const ipRows = useMemo(() => buildIpRows(service, probes), [service, probes])
   const fqdnRows = useMemo(() => buildFqdnRows(service), [service])
-  const nodeRows = useMemo(() => buildNodeRows(nodes), [nodes])
+  const nodeRows = useMemo(() => buildNodeRows(nodes, probes), [nodes, probes])
   const markActive = service.lb_mode === 'failover' || service.lb_mode === 'weighted'
 
   const tabs = TABS.map((entry) => ({

apps/web/src/components/services/service-fqdn-list.tsx

@@ -133,6 +133,7 @@ const VISIBLE_IP_LIMIT = 6
 interface ServiceIpListProps {
   ips: string[]
   ipHealth?: ServiceView['ip_health']
+  healthCheckEnabled?: boolean
   ipEnabled?: Record<string, boolean>
   togglingIp?: string | null
   ipToggleDisabled?: boolean
@@ -151,6 +152,7 @@ interface ServiceIpListProps {
 export function ServiceIpList({
   ips,
   ipHealth = [],
+  healthCheckEnabled = true,
   ipEnabled = {},
   togglingIp = null,
   ipToggleDisabled = false,
@@ -184,6 +186,10 @@ export function ServiceIpList({
       {visible.map((ip) => {
         const health = healthByIp.get(ip)
         const enabled = ipEnabled[ip] !== false
+        const monitored = healthCheckEnabled && enabled
+        const badgeStatus = monitored
+          ? (health?.status ?? 'unknown')
+          : 'disabled'
         return (
           <Item
             key={ip}
@@ -192,7 +198,7 @@ export function ServiceIpList({
           >
             <ItemMedia>
               <HealthCheckBadge
-                status={health?.status ?? 'unknown'}
+                status={badgeStatus}
                 latencyMs={health?.latency_ms}
                 lastCheckedAt={health?.last_checked_at}
                 lastError={health?.last_error}

apps/web/src/components/services/service-unit-card.tsx

@@ -5,6 +5,7 @@ import {
   Repeat2Icon,
   ScaleIcon,
   ServerIcon,
+  UnplugIcon,
   type LucideIcon,
 } from 'lucide-react'
 
@@ -21,6 +22,7 @@ import {
   ServiceIpList,
 } from '@/components/services/service-fqdn-list'
 import { serviceDisplayFqdn, serviceDisplayFqdns } from '@/lib/service-utils'
+import { uniqueIpCount } from '@/lib/failover-events'
 import type { ServiceView } from '@/lib/schemas'
 import { Button } from '@cfdm/ui/components/button'
 import {
@@ -71,11 +73,39 @@ const LB_MODE_META: Record<
   weighted: {
     icon: ScaleIcon,
     className: 'text-info',
-    label: 'Weighted (веса)',
+    label: 'Веса (подмена IP)',
   },
 }
 
-export function LbModeTile({ mode }: { mode: LbMode }) {
+export function LbModeTile({
+  mode,
+  hasPool = true,
+}: {
+  mode: LbMode
+  hasPool?: boolean
+}) {
+  if (!hasPool) {
+    return (
+      <TooltipProvider>
+        <Tooltip>
+          <TooltipTrigger
+            render={
+              <IconTile
+                variant="elevated"
+                size="xs"
+                className="shrink-0 text-muted-foreground"
+                aria-label="без резервирования"
+              />
+            }
+          >
+            <UnplugIcon aria-hidden="true" />
+          </TooltipTrigger>
+          <TooltipContent>без резервирования</TooltipContent>
+        </Tooltip>
+      </TooltipProvider>
+    )
+  }
+
   const meta = LB_MODE_META[mode]
   const Icon = meta.icon
 
@@ -148,7 +178,10 @@ export function ServiceUnitCard({
                   {service.name}
                 </Link>
               </FrameTitle>
-              <LbModeTile mode={service.lb_mode} />
+              <LbModeTile
+                mode={service.lb_mode}
+                hasPool={uniqueIpCount(service.ips) >= 2}
+              />
             </div>
             <div className="flex min-w-0 items-center gap-1">
               <FrameDescription className="min-w-0 truncate font-mono text-xs">
@@ -240,6 +273,10 @@ export function ServiceUnitCard({
           alignWithMenu
           ips={service.ips ?? []}
           ipHealth={service.ip_health ?? []}
+          healthCheckEnabled={
+            service.enabled &&
+            (service.domains ?? []).some((domain) => domain.health_check_enabled)
+          }
           ipEnabled={service.ip_enabled ?? {}}
           ipToggleDisabled={togglingId === service.id}
           togglingIp={togglingIp}

apps/web/src/lib/failover-events.test.ts

@@ -0,0 +1,336 @@
+import { describe, expect, it } from 'vitest'
+
+import {
+  failoverEventCopy,
+  isFailoverEventStatus,
+  mergeFailoverHistory,
+  toFailoverEvents,
+  toIpAliveTransitions,
+  type FailoverBindingPool,
+  type FailoverHealthInput,
+} from '@/lib/failover-events'
+import type { FailoverLogEntry } from '@/lib/schemas'
+import type { HealthLogProbe } from '@/lib/health-log'
+
+function row(
+  overrides: Partial<FailoverHealthInput> & Pick<FailoverHealthInput, 'ip'>,
+): FailoverHealthInput {
+  return {
+    status: 'up',
+    consecutive_failures: 0,
+    last_error: null,
+    last_checked_at: null,
+    ...overrides,
+  }
+}
+
+const mskHip: FailoverBindingPool[] = [
+  {
+    fqdn: 'gt.rkns.top',
+    configured: ['130.49.213.153', '93.115.203.183'],
+    active: ['93.115.203.183'],
+  },
+  {
+    fqdn: 'nsgt.rkns.top',
+    configured: ['130.49.213.153'],
+    active: ['130.49.213.153'],
+  },
+  {
+    fqdn: 'rutg.rkns.top',
+    configured: ['93.115.203.183'],
+    active: ['93.115.203.183'],
+  },
+]
+
+describe('toFailoverEvents', () => {
+  it('инцидент только при binding down, не при unhealthy ноды', () => {
+    expect(isFailoverEventStatus('down')).toBe(true)
+    expect(isFailoverEventStatus('up')).toBe(false)
+    expect(isFailoverEventStatus('degraded')).toBe(false)
+    expect(isFailoverEventStatus('unknown')).toBe(false)
+    expect(isFailoverEventStatus('unhealthy')).toBe(false)
+  })
+
+  it('один IP у сервиса — не инцидент пула', () => {
+    const events = toFailoverEvents(
+      [
+        row({
+          ip: '10.0.0.1',
+          status: 'down',
+          consecutive_failures: 9,
+          last_error: 'fetch failed',
+        }),
+      ],
+      [
+        {
+          fqdn: 'solo.example.com',
+          configured: ['10.0.0.1'],
+          active: [],
+        },
+      ],
+    )
+    expect(events).toEqual([])
+  })
+
+  it('Down без shared pool не событие балансировки', () => {
+    const events = toFailoverEvents([
+      row({
+        ip: '130.49.213.153',
+        status: 'down',
+        consecutive_failures: 9,
+        last_error: 'fetch failed',
+      }),
+    ])
+    expect(events).toEqual([])
+  })
+
+  it('OK вне пула не инцидент (standby)', () => {
+    const events = toFailoverEvents(
+      [
+        row({ ip: '10.0.0.1', status: 'up' }),
+        row({ ip: '130.49.213.153', status: 'up' }),
+      ],
+      [
+        {
+          fqdn: 'pool.example.com',
+          configured: ['10.0.0.1', '130.49.213.153'],
+          active: ['10.0.0.1'],
+        },
+      ],
+    )
+    expect(events).toEqual([])
+  })
+
+  it('unknown и degraded вне пула не инцидент', () => {
+    const events = toFailoverEvents(
+      [
+        row({ ip: '10.0.0.1', status: 'up' }),
+        row({ ip: '10.0.0.2', status: 'unknown' }),
+        row({ ip: '10.0.0.3', status: 'degraded' }),
+      ],
+      [
+        {
+          fqdn: 'pool.example.com',
+          configured: ['10.0.0.1', '10.0.0.2', '10.0.0.3'],
+          active: ['10.0.0.1'],
+        },
+      ],
+    )
+    expect(events).toEqual([])
+  })
+
+  it('Down на доп. FQDN last-resort, снятая с общего пула', () => {
+    const events = toFailoverEvents(
+      [
+        row({
+          ip: '130.49.213.153',
+          status: 'down',
+          consecutive_failures: 9,
+          last_error: 'fetch failed',
+          last_checked_at: '2026-08-20 07:00:00',
+        }),
+        row({ ip: '93.115.203.183', status: 'up' }),
+      ],
+      mskHip,
+    )
+    expect(events).toEqual([
+      {
+        id: '130.49.213.153',
+        address: '130.49.213.153',
+        status: 'down',
+        kind: 'removed',
+        fqdns: ['gt.rkns.top'],
+        consecutiveFailures: 9,
+        lastFailureReason: 'fetch failed',
+        lastCheckAt: '2026-08-20 07:00:00',
+      },
+    ])
+    expect(failoverEventCopy(events[0]!)).toBe('Снята с gt.rkns.top')
+    expect(events[0]?.fqdns).not.toContain('nsgt.rkns.top')
+  })
+
+  it('Down только last-resort на своём FQDN', () => {
+    const events = toFailoverEvents(
+      [row({ ip: '130.49.213.153', status: 'down' })],
+      [
+        {
+          fqdn: 'nsgt.rkns.top',
+          configured: ['130.49.213.153'],
+          active: ['130.49.213.153'],
+        },
+      ],
+    )
+    expect(events).toEqual([])
+  })
+
+  it('down без A-записи на configured FQDN — снятие', () => {
+    const events = toFailoverEvents(
+      [
+        row({
+          ip: '130.49.213.153',
+          status: 'down',
+          consecutive_failures: 9,
+          last_error: 'fetch failed',
+          last_checked_at: '2026-08-20 07:00:00',
+        }),
+      ],
+      [
+        {
+          fqdn: 'pool.example.com',
+          configured: ['10.0.0.1', '130.49.213.153'],
+          active: ['10.0.0.1'],
+        },
+      ],
+    )
+    expect(events).toEqual([
+      {
+        id: '130.49.213.153',
+        address: '130.49.213.153',
+        status: 'down',
+        kind: 'removed',
+        fqdns: ['pool.example.com'],
+        consecutiveFailures: 9,
+        lastFailureReason: 'fetch failed',
+        lastCheckAt: '2026-08-20 07:00:00',
+      },
+    ])
+  })
+})
+
+function probe(
+  overrides: Partial<HealthLogProbe> & Pick<HealthLogProbe, 'id' | 'status' | 'checked_at'>,
+): HealthLogProbe {
+  return {
+    ip: '130.49.213.153',
+    provider: 'local',
+    ok: overrides.status === 'up',
+    latency_ms: 12,
+    colo: null,
+    error: null,
+    ...overrides,
+  }
+}
+
+function dns(
+  overrides: Partial<FailoverLogEntry> & Pick<FailoverLogEntry, 'id' | 'action' | 'created_at'>,
+): FailoverLogEntry {
+  return {
+    service_id: 13,
+    binding_id: 1,
+    fqdn: 'gt.rkns.top',
+    ip: '130.49.213.153',
+    ...overrides,
+  }
+}
+
+describe('toIpAliveTransitions', () => {
+  it('emits leave then return, not the initial up', () => {
+    const transitions = toIpAliveTransitions([
+      probe({ id: 1, status: 'up', checked_at: '2026-08-20T09:00:00Z' }),
+      probe({ id: 2, status: 'up', checked_at: '2026-08-20T09:01:00Z' }),
+      probe({ id: 3, status: 'down', checked_at: '2026-08-20T09:02:00Z' }),
+      probe({ id: 4, status: 'down', checked_at: '2026-08-20T09:03:00Z' }),
+      probe({ id: 5, status: 'up', checked_at: '2026-08-20T09:04:00Z' }),
+    ])
+    expect(transitions).toEqual([
+      { id: 'probe:3', ip: '130.49.213.153', alive: false, at: '2026-08-20T09:02:00Z' },
+      { id: 'probe:5', ip: '130.49.213.153', alive: true, at: '2026-08-20T09:04:00Z' },
+    ])
+  })
+
+  it('any-up across providers: leave only when every source is down', () => {
+    const transitions = toIpAliveTransitions([
+      probe({
+        id: 1,
+        provider: 'local',
+        status: 'up',
+        checked_at: '2026-08-20T09:00:00Z',
+      }),
+      probe({
+        id: 2,
+        provider: 'cloudflare',
+        status: 'down',
+        checked_at: '2026-08-20T09:01:00Z',
+      }),
+      probe({
+        id: 3,
+        provider: 'local',
+        status: 'down',
+        checked_at: '2026-08-20T09:02:00Z',
+      }),
+      probe({
+        id: 4,
+        provider: 'local',
+        status: 'up',
+        checked_at: '2026-08-20T09:03:00Z',
+      }),
+    ])
+    expect(transitions.map((item) => item.id)).toEqual(['probe:3', 'probe:4'])
+  })
+})
+
+describe('mergeFailoverHistory', () => {
+  it('fills leave/return from probes when DNS has only the add', () => {
+    const history = mergeFailoverHistory(
+      [dns({ id: 10, action: 'added', created_at: '2026-08-20 09:26:00' })],
+      [
+        probe({ id: 1, status: 'up', checked_at: '2026-08-20T09:00:00Z' }),
+        probe({ id: 2, status: 'down', checked_at: '2026-08-20T09:10:00Z' }),
+        probe({ id: 3, status: 'up', checked_at: '2026-08-20T09:26:30Z' }),
+      ],
+      mskHip,
+    )
+    expect(history.map((item) => `${item.action}:${item.source}`)).toEqual([
+      'added:dns',
+      'removed:probe',
+    ])
+    expect(history[0]?.copy).toBe('130.49.213.153 добавлена на gt.rkns.top')
+    expect(history[1]?.copy).toBe(
+      '130.49.213.153 вышла из пула (gt.rkns.top)',
+    )
+  })
+
+  it('keeps DNS over a probe return in the same 2-minute window', () => {
+    const history = mergeFailoverHistory(
+      [dns({ id: 10, action: 'added', created_at: '2026-08-20 09:26:00' })],
+      [probe({ id: 3, status: 'up', checked_at: '2026-08-20T09:26:30Z' })],
+      mskHip,
+    )
+    expect(history).toHaveLength(1)
+    expect(history[0]?.source).toBe('dns')
+  })
+
+  it('drops probe leave/return when the service has no shared pool', () => {
+    const history = mergeFailoverHistory(
+      [],
+      [
+        probe({ id: 1, status: 'up', checked_at: '2026-08-20T09:00:00Z' }),
+        probe({ id: 2, status: 'down', checked_at: '2026-08-20T09:10:00Z' }),
+      ],
+      [
+        {
+          fqdn: 'solo.example.com',
+          configured: ['130.49.213.153'],
+          active: ['130.49.213.153'],
+        },
+      ],
+    )
+    expect(history).toEqual([])
+  })
+
+  it('drops dedicated extra-FQDN DNS rows', () => {
+    const history = mergeFailoverHistory(
+      [
+        dns({
+          id: 11,
+          fqdn: 'nsgt.rkns.top',
+          action: 'added',
+          created_at: '2026-08-20 09:26:00',
+        }),
+      ],
+      [],
+      mskHip,
+    )
+    expect(history).toEqual([])
+  })
+})

apps/web/src/lib/failover-events.ts

@@ -0,0 +1,262 @@
+import {
+  bestAliveHealthStatus,
+  probeTime,
+  type HealthLogProbe,
+  type HealthLogStatus,
+} from '@/lib/health-log'
+import type { FailoverLogEntry } from '@/lib/schemas'
+
+export type FailoverEventKind = 'removed' | 'last-resort'
+
+export interface FailoverHistoryItem {
+  id: string
+  ip: string
+  fqdn: string
+  action: 'added' | 'removed'
+  created_at: string
+  copy: string
+  source: 'dns' | 'probe'
+}
+
+export interface FailoverEvent {
+  id: string
+  address: string
+  status: string
+  kind: FailoverEventKind
+  fqdns: string[]
+  consecutiveFailures: number
+  lastFailureReason: string | null
+  lastCheckAt?: string | null
+}
+
+/** Binding IP health — тот же контур, что таблица активов. */
+export interface FailoverHealthInput {
+  ip: string
+  status: string
+  consecutive_failures?: number
+  last_error?: string | null
+  last_checked_at?: string | null
+}
+
+export interface FailoverBindingPool {
+  fqdn: string
+  configured: readonly string[]
+  active: readonly string[]
+}
+
+/** Unique A targets. Duplicates of the same IP are not a pool. */
+export function uniqueIpCount(ips: readonly string[]): number {
+  return new Set(ips.filter(Boolean)).size
+}
+
+/** Shared pool FQDN — two or more unique A targets. */
+export function isSharedPoolBinding(binding: FailoverBindingPool): boolean {
+  return uniqueIpCount(binding.configured) >= 2
+}
+
+export function hasSharedPool(
+  bindings: readonly FailoverBindingPool[],
+): boolean {
+  return bindings.some(isSharedPoolBinding)
+}
+
+/** Инцидент только при down. up / degraded / unknown — не вывод из пула. */
+export function isFailoverEventStatus(status: string): boolean {
+  return status === 'down'
+}
+
+export function failoverEventCopy(event: FailoverEvent): string {
+  if (event.kind === 'removed') {
+    return event.fqdns.length > 0
+      ? `Снята с ${event.fqdns.join(', ')}`
+      : 'Снята с DNS'
+  }
+  return event.fqdns.length > 0
+    ? `Down, в A-записях last-resort на ${event.fqdns.join(', ')}`
+    : 'Down, в A-записях last-resort'
+}
+
+/**
+ * Инциденты пула только если у сервиса есть shared FQDN (2+ уникальных IP).
+ * Один IP — не балансировка и не вывод из пула; Down смотрит health-монитор.
+ */
+export function toFailoverEvents(
+  ipHealth: readonly FailoverHealthInput[],
+  bindings: readonly FailoverBindingPool[] = [],
+): FailoverEvent[] {
+  if (!hasSharedPool(bindings)) return []
+  return ipHealth.filter((row) => isFailoverEventStatus(row.status)).map((row) => {
+    const removedFqdns: string[] = []
+    const lastResortFqdns: string[] = []
+    for (const binding of bindings) {
+      if (!isSharedPoolBinding(binding)) continue
+      const configured = binding.configured.includes(row.ip)
+      const active = binding.active.includes(row.ip)
+      if (configured && !active) removedFqdns.push(binding.fqdn)
+      else if (configured && active) lastResortFqdns.push(binding.fqdn)
+    }
+    const kind: FailoverEventKind =
+      removedFqdns.length > 0 ? 'removed' : 'last-resort'
+    return {
+      id: row.ip,
+      address: row.ip,
+      status: row.status,
+      kind,
+      fqdns: kind === 'removed' ? removedFqdns : lastResortFqdns,
+      consecutiveFailures: row.consecutive_failures ?? 0,
+      lastFailureReason: row.last_error ?? null,
+      lastCheckAt: row.last_checked_at ?? null,
+    }
+  })
+}
+
+const DEDUPE_WINDOW_MS = 2 * 60 * 1000
+
+function fqdnsForIp(
+  ip: string,
+  bindings: readonly FailoverBindingPool[],
+): string[] {
+  return bindings
+    .filter((binding) => isSharedPoolBinding(binding) && binding.configured.includes(ip))
+    .map((binding) => binding.fqdn)
+}
+
+function isPoolFqdn(
+  fqdn: string,
+  bindings: readonly FailoverBindingPool[],
+): boolean {
+  const binding = bindings.find((item) => item.fqdn === fqdn)
+  if (!binding) return true
+  return isSharedPoolBinding(binding)
+}
+
+function fqdnLabel(fqdns: string[]): string {
+  return fqdns.join(', ') || 'пул'
+}
+
+function isAliveStatus(status: HealthLogStatus): boolean {
+  return status === 'up' || status === 'degraded'
+}
+
+/**
+ * Per-IP any-up flips: down → вышла из пула, up после down → вернулась.
+ * Initial state is not an event.
+ */
+export function toIpAliveTransitions(
+  probes: readonly HealthLogProbe[],
+): Array<{ id: string; ip: string; alive: boolean; at: string }> {
+  const byIp = new Map<string, HealthLogProbe[]>()
+  for (const item of probes) {
+    const list = byIp.get(item.ip)
+    if (list) list.push(item)
+    else byIp.set(item.ip, [item])
+  }
+
+  const out: Array<{ id: string; ip: string; alive: boolean; at: string }> = []
+  for (const [ip, list] of byIp) {
+    list.sort(
+      (a, b) => probeTime(a.checked_at) - probeTime(b.checked_at) || a.id - b.id,
+    )
+    const latestByProvider = new Map<string, HealthLogProbe>()
+    let prevAlive: boolean | undefined
+    for (const probe of list) {
+      latestByProvider.set(probe.provider, probe)
+      const status = bestAliveHealthStatus(
+        [...latestByProvider.values()].map((item) => item.status),
+      )
+      if (status === 'unknown') continue
+      const alive = isAliveStatus(status)
+      if (prevAlive !== undefined && alive !== prevAlive) {
+        out.push({
+          id: `probe:${probe.id}`,
+          ip,
+          alive,
+          at: probe.checked_at,
+        })
+      }
+      prevAlive = alive
+    }
+  }
+  return out
+}
+
+function dnsHistoryItem(item: FailoverLogEntry): FailoverHistoryItem {
+  return {
+    id: `dns:${item.id}`,
+    ip: item.ip,
+    fqdn: item.fqdn,
+    action: item.action,
+    created_at: item.created_at,
+    copy:
+      item.action === 'removed'
+        ? `${item.ip} убрана с ${item.fqdn}`
+        : `${item.ip} добавлена на ${item.fqdn}`,
+    source: 'dns',
+  }
+}
+
+function probeHistoryItem(
+  transition: { id: string; ip: string; alive: boolean; at: string },
+  bindings: readonly FailoverBindingPool[],
+): FailoverHistoryItem {
+  const fqdns = fqdnsForIp(transition.ip, bindings)
+  const fqdn = fqdnLabel(fqdns)
+  const action = transition.alive ? 'added' : 'removed'
+  return {
+    id: transition.id,
+    ip: transition.ip,
+    fqdn,
+    action,
+    created_at: transition.at,
+    copy:
+      action === 'removed'
+        ? `${transition.ip} вышла из пула (${fqdn})`
+        : `${transition.ip} вернулась в пул (${fqdn})`,
+    source: 'probe',
+  }
+}
+
+function eventTime(value: string): number {
+  return probeTime(value)
+}
+
+/**
+ * DNS add/remove + health leave/return, newest first.
+ * Same IP+action within 2 minutes: keep the DNS row (it has a concrete FQDN).
+ */
+export function mergeFailoverHistory(
+  dns: readonly FailoverLogEntry[],
+  probes: readonly HealthLogProbe[],
+  bindings: readonly FailoverBindingPool[] = [],
+): FailoverHistoryItem[] {
+  const merged = [
+    ...dns
+      .filter((item) => isPoolFqdn(item.fqdn, bindings))
+      .map(dnsHistoryItem),
+    ...toIpAliveTransitions(probes)
+      .filter((transition) => fqdnsForIp(transition.ip, bindings).length > 0)
+      .map((transition) => probeHistoryItem(transition, bindings)),
+  ]
+  merged.sort(
+    (a, b) => eventTime(b.created_at) - eventTime(a.created_at) || a.id.localeCompare(b.id),
+  )
+
+  const kept: FailoverHistoryItem[] = []
+  for (const item of merged) {
+    const duplicate = kept.find(
+      (other) =>
+        other.ip === item.ip &&
+        other.action === item.action &&
+        Math.abs(eventTime(other.created_at) - eventTime(item.created_at)) <=
+          DEDUPE_WINDOW_MS,
+    )
+    if (!duplicate) {
+      kept.push(item)
+      continue
+    }
+    if (duplicate.source === 'probe' && item.source === 'dns') {
+      kept[kept.indexOf(duplicate)] = item
+    }
+  }
+  return kept
+}

apps/web/src/lib/health-log.test.ts

@@ -1,9 +1,13 @@
 import { describe, expect, it } from 'vitest'
 
 import {
+  bestAliveHealthStatus,
   collapseStatusChanges,
   enabledHealthProviders,
+  latestHealthByIp,
   providerHealthStatuses,
+  resolveIpDisplayHealth,
+  resolveServiceDisplayHealth,
   worstHealthStatus,
   type HealthLogProbe,
 } from '@/lib/health-log'
@@ -73,7 +77,7 @@ describe('enabledHealthProviders', () => {
 })
 
 describe('providerHealthStatuses', () => {
-  it('uses worst latest-per-ip status', () => {
+  it('uses any-up among latest-per-ip statuses', () => {
     const items = [
       probe({ id: 1, ip: '1.1.1.1', status: 'up', checked_at: '2026-01-01T00:02:00Z' }),
       probe({ id: 2, ip: '2.2.2.2', status: 'down', checked_at: '2026-01-01T00:01:00Z' }),
@@ -84,7 +88,7 @@ describe('providerHealthStatuses', () => {
         checked_at: '2026-01-01T00:00:00Z',
       }),
     ]
-    expect(providerHealthStatuses(items, ['local']).local).toBe('down')
+    expect(providerHealthStatuses(items, ['local']).local).toBe('up')
   })
 })
 
@@ -95,3 +99,70 @@ describe('worstHealthStatus', () => {
     expect(worstHealthStatus([])).toBe('unknown')
   })
 })
+
+describe('bestAliveHealthStatus', () => {
+  it('is up if any IP is up', () => {
+    expect(bestAliveHealthStatus(['up', 'down'])).toBe('up')
+    expect(bestAliveHealthStatus(['degraded', 'down'])).toBe('degraded')
+    expect(bestAliveHealthStatus(['down'])).toBe('down')
+    expect(bestAliveHealthStatus([])).toBe('unknown')
+  })
+})
+
+describe('latestHealthByIp', () => {
+  it('any-up among latest-per-provider probes', () => {
+    const items = [
+      probe({
+        id: 1,
+        ip: '130.49.213.153',
+        provider: 'local',
+        status: 'up',
+        checked_at: '2026-08-20T09:00:00Z',
+      }),
+      probe({
+        id: 2,
+        ip: '130.49.213.153',
+        provider: 'cloudflare',
+        status: 'down',
+        checked_at: '2026-08-20T09:00:01Z',
+      }),
+      probe({
+        id: 3,
+        ip: '93.115.203.183',
+        status: 'unknown',
+        checked_at: '2026-08-20T08:59:00Z',
+      }),
+    ]
+    const byIp = latestHealthByIp(items)
+    expect(byIp.get('130.49.213.153')?.status).toBe('up')
+    expect(byIp.get('93.115.203.183')?.status).toBe('unknown')
+  })
+})
+
+describe('resolveIpDisplayHealth', () => {
+  it('prefers a live probe over stored unknown', () => {
+    expect(resolveIpDisplayHealth('unknown', 'up')).toBe('up')
+    expect(resolveIpDisplayHealth('down', 'up')).toBe('up')
+    expect(resolveIpDisplayHealth('up', 'unknown')).toBe('up')
+    expect(resolveIpDisplayHealth('unknown', undefined)).toBe('unknown')
+  })
+})
+
+describe('resolveServiceDisplayHealth', () => {
+  it('shows OK when live probes recovered and stored is still unknown', () => {
+    expect(
+      resolveServiceDisplayHealth(
+        'unknown',
+        [{ ip: '2.59.161.102', status: 'unknown' }],
+        [
+          probe({
+            id: 1,
+            ip: '2.59.161.102',
+            status: 'up',
+            checked_at: '2026-08-20T18:00:00Z',
+          }),
+        ],
+      ),
+    ).toBe('up')
+  })
+})

apps/web/src/lib/health-log.ts

@@ -98,7 +98,18 @@ export function worstHealthStatus(statuses: readonly HealthLogStatus[]): HealthL
   )
 }
 
-/** Latest probe per IP for a provider, then worst among those IPs. */
+/** Any up → up; else degraded, then down, then unknown. */
+export function bestAliveHealthStatus(
+  statuses: readonly HealthLogStatus[],
+): HealthLogStatus {
+  if (statuses.length === 0) return 'unknown'
+  if (statuses.some((status) => status === 'up')) return 'up'
+  if (statuses.some((status) => status === 'degraded')) return 'degraded'
+  if (statuses.some((status) => status === 'down')) return 'down'
+  return 'unknown'
+}
+
+/** Latest probe per IP for a provider, then any-up among those IPs. */
 export function providerHealthStatuses(
   items: readonly HealthLogProbe[],
   providers: readonly HealthCheckProvider[],
@@ -120,8 +131,85 @@ export function providerHealthStatuses(
     const statuses = [...latestByIp.values()]
       .filter((item) => item.provider === provider)
       .map((item) => item.status)
-    result[provider] = worstHealthStatus(statuses)
+    result[provider] = bestAliveHealthStatus(statuses)
   }
 
   return result
 }
+
+export interface IpDisplayHealth {
+  status: HealthLogStatus
+  latency_ms: number | null
+  last_checked_at: string
+  last_error: string | null
+  colo: string | null
+  provider: HealthCheckProvider
+}
+
+/**
+ * Latest probe per provider+IP, then any-up among those providers.
+ * Used by the IP table so hysteresis `unknown` in ip_health does not hide a live OK.
+ */
+export function latestHealthByIp(
+  items: readonly HealthLogProbe[],
+): Map<string, IpDisplayHealth> {
+  const latest = new Map<string, HealthLogProbe>()
+  const sorted = [...items].sort(
+    (a, b) => probeTime(b.checked_at) - probeTime(a.checked_at) || b.id - a.id,
+  )
+  for (const item of sorted) {
+    const key = `${item.provider}\0${item.ip}`
+    if (!latest.has(key)) latest.set(key, item)
+  }
+
+  const byIp = new Map<string, HealthLogProbe[]>()
+  for (const item of latest.values()) {
+    const list = byIp.get(item.ip)
+    if (list) list.push(item)
+    else byIp.set(item.ip, [item])
+  }
+
+  const result = new Map<string, IpDisplayHealth>()
+  for (const [ip, probes] of byIp) {
+    const status = bestAliveHealthStatus(probes.map((probe) => probe.status))
+    const preferred =
+      probes.find((probe) => probe.status === status) ?? probes[0]!
+    result.set(ip, {
+      status,
+      latency_ms: preferred.latency_ms,
+      last_checked_at: preferred.checked_at,
+      last_error: preferred.error,
+      colo: preferred.colo,
+      provider: preferred.provider,
+    })
+  }
+  return result
+}
+
+/** Prefer a concrete live probe over stored hysteresis `unknown`. */
+export function resolveIpDisplayHealth(
+  stored: HealthLogStatus | undefined,
+  live: HealthLogStatus | undefined,
+): HealthLogStatus {
+  if (live && live !== 'unknown') return live
+  return stored ?? live ?? 'unknown'
+}
+
+/** Service KPI / card: any-up of per-IP overlay (live probes beat hysteresis). */
+export function resolveServiceDisplayHealth(
+  stored: HealthLogStatus | undefined,
+  ipHealth: readonly { ip: string; status: string }[],
+  probes: readonly HealthLogProbe[] = [],
+): HealthLogStatus {
+  const liveByIp = latestHealthByIp(probes)
+  const statuses =
+    ipHealth.length > 0
+      ? ipHealth.map((row) =>
+          resolveIpDisplayHealth(
+            row.status as HealthLogStatus,
+            liveByIp.get(row.ip)?.status,
+          ),
+        )
+      : [...liveByIp.values()].map((row) => row.status)
+  return resolveIpDisplayHealth(stored, bestAliveHealthStatus(statuses))
+}

apps/web/src/lib/schemas.ts

@@ -84,6 +84,7 @@ export const serviceDomainBindingSchema = z
     health_check_aggregate: z.enum(['any', 'all', 'majority']).catch('majority'),
     cert_monitoring: z.enum(['auto', 'required', 'skipped']).default('auto'),
     sync_status: z.string().nullable().default(null),
+    active_ips: z.array(z.string()).default([]),
   })
   .transform((binding) => ({
     ...binding,
@@ -125,6 +126,18 @@ export const healthProbeLogSchema = z.object({
   checked_at: z.string(),
 })
 
+export const failoverLogSchema = z.object({
+  id: z.number(),
+  service_id: z.number(),
+  binding_id: z.number(),
+  fqdn: z.string(),
+  ip: z.string(),
+  action: z.enum(['added', 'removed']),
+  created_at: z.string(),
+})
+
+export type FailoverLogEntry = z.infer<typeof failoverLogSchema>
+
 export const serviceViewSchema = serviceSchema.extend({
   subdomain: z.string().default(''),
   enabled: z.coerce.boolean().default(false),

apps/web/src/lib/service-address.test.ts

@@ -3,9 +3,12 @@ import { describe, expect, it } from 'vitest'
 import {
   DEFAULT_BINDING_HEALTH,
   addAddressNode,
+  addCommonFqdn,
+  addExtraFqdn,
   emptyAddressBlock,
   emptyBindingDraft,
   hydrateAddressBlock,
+  patchAddressIpMeta,
   removeAddressNode,
   toAddressBindings,
   toDomainsPayload,
@@ -41,15 +44,15 @@ describe('hydrateAddressBlock', () => {
 
     const state = hydrateAddressBlock(drafts, ['93.115.203.183', '185.244.181.61'])
 
-    expect(state.commonFqdn).toBe('rutg.rkns.top')
+    expect(state.commonFqdns).toEqual(['rutg.rkns.top'])
     expect(state.nodes).toEqual([
-      { ip: '93.115.203.183', extraFqdn: 'msk.rutg.rkns.top' },
-      { ip: '185.244.181.61', extraFqdn: '' },
+      { ip: '93.115.203.183', extraFqdns: ['msk.rutg.rkns.top'] },
+      { ip: '185.244.181.61', extraFqdns: [] },
     ])
-    expect(state.otherBindings).toEqual([])
+    expect(state.preservedBindings).toEqual([])
   })
 
-  it('не схлопывает CNAME и A на несколько IP', () => {
+  it('кладёт A на весь пул в commonFqdns, CNAME — в preserved', () => {
     const cname: ServiceBindingDraft = {
       ...emptyBindingDraft('alias.rkns.top'),
       record_type: 'CNAME',
@@ -63,14 +66,12 @@ describe('hydrateAddressBlock', () => {
 
     const state = hydrateAddressBlock(drafts, ['1.1.1.1', '2.2.2.2'])
 
-    expect(state.nodes.every((node) => node.extraFqdn === '')).toBe(true)
-    expect(state.otherBindings.map((item) => item.fqdn)).toEqual([
-      'both.rkns.top',
-      'alias.rkns.top',
-    ])
+    expect(state.commonFqdns).toEqual(['rutg.rkns.top', 'both.rkns.top'])
+    expect(state.nodes.every((node) => node.extraFqdns.length === 0)).toBe(true)
+    expect(state.preservedBindings.map((item) => item.fqdn)).toEqual(['alias.rkns.top'])
   })
 
-  it('кладёт extra A с IP вне пула в otherBindings', () => {
+  it('кладёт extra A с IP вне пула в preservedBindings', () => {
     const drafts = [
       aRecord('gw.example.com', ['10.0.0.1']),
       aRecord('edge.example.com', ['8.8.8.8']),
@@ -78,14 +79,109 @@ describe('hydrateAddressBlock', () => {
 
     const state = hydrateAddressBlock(drafts, ['10.0.0.1'])
 
-    expect(state.nodes).toEqual([{ ip: '10.0.0.1', extraFqdn: '' }])
-    expect(state.otherBindings).toHaveLength(1)
-    expect(state.otherBindings[0]?.fqdn).toBe('edge.example.com')
+    expect(state.nodes).toEqual([{ ip: '10.0.0.1', extraFqdns: [] }])
+    expect(state.commonFqdns).toEqual(['gw.example.com'])
+    expect(state.preservedBindings).toHaveLength(1)
+    expect(state.preservedBindings[0]?.fqdn).toBe('edge.example.com')
+  })
+
+  it('при одном IP пула отделяет второй A в extraFqdn узла', () => {
+    const drafts = [
+      aRecord('dns.shnt.top', ['130.49.213.176']),
+      aRecord('ndns.shnt.top', ['130.49.213.176']),
+    ]
+
+    const state = hydrateAddressBlock(drafts, ['130.49.213.176'])
+
+    expect(state.commonFqdns).toEqual(['dns.shnt.top'])
+    expect(state.nodes).toEqual([
+      { ip: '130.49.213.176', extraFqdns: ['ndns.shnt.top'] },
+    ])
+    expect(state.preservedBindings).toEqual([])
+  })
+
+  it('кладёт несколько extra A на один IP в extraFqdns, включая wildcard', () => {
+    const drafts = [
+      aRecord('dns.shnt.top', ['130.49.213.176']),
+      aRecord('ndns.shnt.top', ['130.49.213.176']),
+      aRecord('*.mdns.shnt.top', ['130.49.213.176']),
+    ]
+
+    const state = hydrateAddressBlock(drafts, ['130.49.213.176'])
+
+    expect(state.commonFqdns).toEqual(['dns.shnt.top'])
+    expect(state.nodes).toEqual([
+      {
+        ip: '130.49.213.176',
+        extraFqdns: ['ndns.shnt.top', '*.mdns.shnt.top'],
+      },
+    ])
+    expect(state.preservedBindings).toEqual([])
+  })
+
+  it('поднимает веса и приоритеты с общего FQDN', () => {
+    const drafts = [
+      aRecord('gt.rkns.top', ['130.49.213.153', '93.115.203.183'], {
+        target_ip_weights: { '130.49.213.153': 3, '93.115.203.183': 1 },
+        target_ip_priorities: { '130.49.213.153': 2, '93.115.203.183': 1 },
+      }),
+      aRecord('nsgt.rkns.top', ['130.49.213.153']),
+    ]
+    const state = hydrateAddressBlock(drafts, ['130.49.213.153', '93.115.203.183'])
+    expect(state.target_ip_weights).toEqual({
+      '130.49.213.153': 3,
+      '93.115.203.183': 1,
+    })
+    expect(state.target_ip_priorities).toEqual({
+      '130.49.213.153': 2,
+      '93.115.203.183': 1,
+    })
+    expect(state.nodes[0]?.extraFqdns).toEqual(['nsgt.rkns.top'])
+  })
+
+  it('лечит урезанный общий FQDN (legacy toggle) как common, не preserved', () => {
+    const pool = ['130.49.213.153', '130.49.213.176', '93.115.203.183']
+    const drafts = [
+      // corrupted common — missing one pool IP
+      aRecord('gw.pngs.top', ['130.49.213.153', '130.49.213.176']),
+      aRecord('gt.rkns.top', pool),
+      aRecord('nsgt.rkns.top', ['130.49.213.176']),
+      aRecord('rutg.rkns.top', ['93.115.203.183']),
+    ]
+    const state = hydrateAddressBlock(drafts, pool)
+
+    expect(state.commonFqdns).toEqual(['gw.pngs.top', 'gt.rkns.top'])
+    expect(state.nodes).toEqual([
+      { ip: '130.49.213.153', extraFqdns: [] },
+      { ip: '130.49.213.176', extraFqdns: ['nsgt.rkns.top'] },
+      { ip: '93.115.203.183', extraFqdns: ['rutg.rkns.top'] },
+    ])
+    expect(state.preservedBindings).toEqual([])
+
+    const payload = toDomainsPayload(state, primaryMeta)
+    expect(payload.map((item) => item.fqdn)).toEqual([
+      'gw.pngs.top',
+      'gt.rkns.top',
+      'nsgt.rkns.top',
+      'rutg.rkns.top',
+    ])
+    expect(payload[0]?.target_ips).toEqual(pool)
+    expect(new Set(payload.map((item) => item.fqdn.toLowerCase())).size).toBe(4)
+  })
+
+  it('не дублирует FQDN при повторном binding в drafts', () => {
+    const pool = ['1.1.1.1', '2.2.2.2']
+    const state = hydrateAddressBlock(
+      [aRecord('gw.example.com', ['1.1.1.1']), aRecord('gw.example.com', pool)],
+      pool,
+    )
+    expect(state.commonFqdns).toEqual(['gw.example.com'])
+    expect(state.nodes.every((node) => node.extraFqdns.length === 0)).toBe(true)
   })
 })
 
 describe('toDomainsPayload', () => {
-  it('собирает primary на весь пул и extra binding на один IP', () => {
+  it('собирает каждый common на весь пул и extra binding на один IP', () => {
     const drafts = [
       aRecord('rutg.rkns.top', ['93.115.203.183', '185.244.181.61']),
       aRecord('msk.rutg.rkns.top', ['93.115.203.183']),
@@ -107,65 +203,145 @@ describe('toDomainsPayload', () => {
     ])
   })
 
-  it('круг hydrate → payload → hydrate сохраняет extra FQDN', () => {
+  it('круг hydrate → payload → hydrate сохраняет два common и extra FQDN', () => {
     const drafts = [
-      aRecord('rutg.rkns.top', ['93.115.203.183', '185.244.181.61']),
-      aRecord('msk.rutg.rkns.top', ['93.115.203.183']),
+      aRecord('gt.rkns.top', ['93.115.203.183', '185.244.181.61']),
+      aRecord('msk.rkns.top', ['93.115.203.183', '185.244.181.61']),
+      aRecord('nsgt.rkns.top', ['93.115.203.183']),
     ]
     const first = hydrateAddressBlock(drafts, ['93.115.203.183', '185.244.181.61'])
+    expect(first.commonFqdns).toEqual(['gt.rkns.top', 'msk.rkns.top'])
     const rebound = toAddressBindings(first, primaryMeta)
     const second = hydrateAddressBlock(rebound, rebound[0]?.target_ips ?? [])
 
-    expect(second.commonFqdn).toBe(first.commonFqdn)
+    expect(second.commonFqdns).toEqual(first.commonFqdns)
     expect(second.nodes).toEqual(first.nodes)
-    expect(second.otherBindings).toEqual([])
+    expect(second.preservedBindings).toEqual([])
+  })
+
+  it('круг hydrate → payload → hydrate сохраняет extra FQDN при одном IP', () => {
+    const drafts = [
+      aRecord('dns.shnt.top', ['130.49.213.176']),
+      aRecord('ndns.shnt.top', ['130.49.213.176']),
+    ]
+    const first = hydrateAddressBlock(drafts, ['130.49.213.176'])
+    expect(first.commonFqdns).toEqual(['dns.shnt.top'])
+    expect(first.nodes).toEqual([
+      { ip: '130.49.213.176', extraFqdns: ['ndns.shnt.top'] },
+    ])
+    const rebound = toAddressBindings(first, primaryMeta)
+    const second = hydrateAddressBlock(rebound, ['130.49.213.176'])
+
+    expect(second.commonFqdns).toEqual(first.commonFqdns)
+    expect(second.nodes).toEqual(first.nodes)
+    expect(second.preservedBindings).toEqual([])
+  })
+
+  it('круг hydrate → payload → hydrate сохраняет несколько extra и wildcard при одном IP', () => {
+    const drafts = [
+      aRecord('dns.shnt.top', ['130.49.213.176']),
+      aRecord('ndns.shnt.top', ['130.49.213.176']),
+      aRecord('*.mdns.shnt.top', ['130.49.213.176']),
+    ]
+    const first = hydrateAddressBlock(drafts, ['130.49.213.176'])
+    expect(first.nodes[0]?.extraFqdns).toEqual(['ndns.shnt.top', '*.mdns.shnt.top'])
+    const rebound = toAddressBindings(first, primaryMeta)
+    const second = hydrateAddressBlock(rebound, ['130.49.213.176'])
+
+    expect(second.commonFqdns).toEqual(first.commonFqdns)
+    expect(second.nodes).toEqual(first.nodes)
+    expect(second.preservedBindings).toEqual([])
   })
 })
 
 describe('removeAddressNode', () => {
-  it('удаляет extra FQDN узла и IP из other A-bindings', () => {
+  it('удаляет extra FQDN узла и IP из preserved A-bindings', () => {
     const state = hydrateAddressBlock(
       [
         aRecord('gw.example.com', ['10.0.0.1', '10.0.0.2']),
         aRecord('msk.example.com', ['10.0.0.1']),
-        aRecord('pair.example.com', ['10.0.0.1', '10.0.0.2']),
+        aRecord('edge.example.com', ['9.9.9.9', '10.0.0.1']),
       ],
       ['10.0.0.1', '10.0.0.2'],
     )
 
     const next = removeAddressNode(state, '10.0.0.1')
 
-    expect(next.nodes).toEqual([{ ip: '10.0.0.2', extraFqdn: '' }])
-    expect(next.otherBindings).toHaveLength(1)
-    expect(next.otherBindings[0]?.target_ips).toEqual(['10.0.0.2'])
+    expect(next.nodes).toEqual([{ ip: '10.0.0.2', extraFqdns: [] }])
+    expect(next.preservedBindings).toHaveLength(1)
+    expect(next.preservedBindings[0]?.target_ips).toEqual(['9.9.9.9'])
   })
 })
 
-describe('addAddressNode', () => {
+describe('addAddressNode / addCommonFqdn', () => {
   it('не добавляет дубликат IP', () => {
     const withIp = addAddressNode(
-      { ...emptyAddressBlock(), nodes: [{ ip: '1.1.1.1', extraFqdn: '' }] },
+      { ...emptyAddressBlock(), nodes: [{ ip: '1.1.1.1', extraFqdns: [] }] },
       '1.1.1.1',
     )
     expect(withIp.nodes).toHaveLength(1)
   })
+
+  it('не добавляет дубликат common FQDN', () => {
+    const state = addCommonFqdn(
+      { ...emptyAddressBlock(), commonFqdns: ['gt.rkns.top'] },
+      'GT.rkns.top',
+    )
+    expect(state.commonFqdns).toEqual(['gt.rkns.top'])
+  })
+
+  it('добавляет extra FQDN к IP и отклоняет дубликат', () => {
+    const withIp = addAddressNode(emptyAddressBlock(), '1.1.1.1')
+    const withExtra = addExtraFqdn(withIp, '1.1.1.1', 'mdns.shnt.top')
+    expect(withExtra.nodes[0]?.extraFqdns).toEqual(['mdns.shnt.top'])
+    expect(addExtraFqdn(withExtra, '1.1.1.1', 'MDNS.shnt.top')).toBe(withExtra)
+  })
 })
 
-describe('CNAME / otherBindings', () => {
-  it('сохраняет CNAME в otherBindings при круге hydrate → payload', () => {
+describe('patchAddressIpMeta', () => {
+  it('меняет вес одного IP и не трогает extraFqdns', () => {
+    const state = {
+      ...addAddressNode(addAddressNode(emptyAddressBlock(), '1.1.1.1'), '2.2.2.2'),
+      nodes: [
+        { ip: '1.1.1.1', extraFqdns: ['msk.example.com'] },
+        { ip: '2.2.2.2', extraFqdns: [] },
+      ],
+    }
+    const next = patchAddressIpMeta(state, '1.1.1.1', { weight: 7 })
+    expect(next.target_ip_weights).toEqual({ '1.1.1.1': 7, '2.2.2.2': 1 })
+    expect(next.target_ip_priorities).toEqual(state.target_ip_priorities)
+    expect(next.nodes).toEqual(state.nodes)
+  })
+
+  it('clamp веса и приоритета в 1–100', () => {
+    const state = addAddressNode(emptyAddressBlock(), '10.0.0.1')
+    expect(patchAddressIpMeta(state, '10.0.0.1', { weight: 0 }).target_ip_weights['10.0.0.1']).toBe(1)
+    expect(
+      patchAddressIpMeta(state, '10.0.0.1', { priority: 999 }).target_ip_priorities['10.0.0.1'],
+    ).toBe(100)
+  })
+
+  it('игнорирует IP вне пула', () => {
+    const state = addAddressNode(emptyAddressBlock(), '10.0.0.1')
+    expect(patchAddressIpMeta(state, '8.8.8.8', { weight: 5 })).toBe(state)
+  })
+})
+
+describe('CNAME / preservedBindings', () => {
+  it('сохраняет CNAME в preserved при круге hydrate → payload', () => {
     const cname: ServiceBindingDraft = {
       ...emptyBindingDraft('alias.rkns.top'),
       record_type: 'CNAME',
       target_cname: 'rutg.rkns.top',
     }
     const drafts = [
-      aRecord('rutg.rkns.top', ['1.1.1.1']),
+      aRecord('rutg.rkns.top', ['1.1.1.1', '2.2.2.2']),
       aRecord('msk.rkns.top', ['1.1.1.1']),
       cname,
     ]
-    const state = hydrateAddressBlock(drafts, ['1.1.1.1'])
-    expect(state.nodes[0]?.extraFqdn).toBe('msk.rkns.top')
-    expect(state.otherBindings).toHaveLength(1)
+    const state = hydrateAddressBlock(drafts, ['1.1.1.1', '2.2.2.2'])
+    expect(state.nodes[0]?.extraFqdns).toEqual(['msk.rkns.top'])
+    expect(state.preservedBindings).toHaveLength(1)
 
     const payload = toDomainsPayload(state, primaryMeta)
     expect(payload.map((item) => item.fqdn)).toEqual([

apps/web/src/lib/service-address.ts

@@ -33,13 +33,13 @@ export interface ServiceBindingDraft {
 
 export interface AddressNode {
   ip: string
-  extraFqdn: string
+  extraFqdns: string[]
 }
 
 export interface AddressBlockState {
-  commonFqdn: string
+  commonFqdns: string[]
   nodes: AddressNode[]
-  otherBindings: ServiceBindingDraft[]
+  preservedBindings: ServiceBindingDraft[]
   target_ip_weights: Record<string, number>
   target_ip_priorities: Record<string, number>
 }
@@ -78,31 +78,14 @@ export function emptyBindingDraft(fqdn = ''): ServiceBindingDraft {
 
 export function emptyAddressBlock(): AddressBlockState {
   return {
-    commonFqdn: '',
+    commonFqdns: [],
     nodes: [],
-    otherBindings: [],
+    preservedBindings: [],
     target_ip_weights: {},
     target_ip_priorities: {},
   }
 }
 
-export function withPoolIps(
-  draft: ServiceBindingDraft,
-  pool: string[],
-): ServiceBindingDraft {
-  if (draft.record_type !== 'A' || draft.target_ips.length > 0 || pool.length === 0) {
-    return draft
-  }
-  return {
-    ...draft,
-    target_ips: pool,
-    target_ip_weights: Object.fromEntries(pool.map((ip) => [ip, draft.target_ip_weights[ip] ?? 1])),
-    target_ip_priorities: Object.fromEntries(
-      pool.map((ip) => [ip, draft.target_ip_priorities[ip] ?? 1]),
-    ),
-  }
-}
-
 function uniqueIps(...lists: string[][]): string[] {
   const seen = new Set<string>()
   const out: string[] = []
@@ -123,6 +106,17 @@ function omitKey(record: Record<string, number>, key: string): Record<string, nu
   return next
 }
 
+function sameIpSet(left: string[], right: string[]): boolean {
+  if (left.length === 0 || left.length !== right.length) return false
+  const set = new Set(left.map((ip) => ip.trim()).filter(Boolean))
+  if (set.size !== left.length) return false
+  return right.every((ip) => set.has(ip.trim()))
+}
+
+function fqdnKey(value: string): string {
+  return value.trim().toLowerCase()
+}
+
 export function toBindingDrafts(service: ServiceView): ServiceBindingDraft[] {
   return (service.domains ?? []).map((binding) => ({
     fqdn: bindingToFqdn(binding),
@@ -151,49 +145,138 @@ export function toBindingDrafts(service: ServiceView): ServiceBindingDraft[] {
   }))
 }
 
-function canCollapseToNode(
-  extra: ServiceBindingDraft,
-  pool: Set<string>,
-  claimed: Set<string>,
-): string | null {
-  if (extra.record_type !== 'A') return null
-  if (extra.target_ips.length !== 1) return null
-  const ip = extra.target_ips[0]?.trim() ?? ''
-  if (!ip || !pool.has(ip) || claimed.has(ip)) return null
-  if (!extra.fqdn.trim()) return null
-  return ip
+function isFullPoolA(draft: ServiceBindingDraft, pool: string[]): boolean {
+  return draft.record_type === 'A' && sameIpSet(draft.target_ips, pool)
+}
+
+/** UI contract: common = 2+ IPs all in pool (full or corrupted subset after old toggles). */
+function isCommonPoolA(draft: ServiceBindingDraft, poolSet: Set<string>): boolean {
+  if (draft.record_type !== 'A') return false
+  const ips = draft.target_ips.map((ip) => ip.trim()).filter(Boolean)
+  if (ips.length < 2) return false
+  return ips.every((ip) => poolSet.has(ip))
+}
+
+function takeAsCommon(
+  draft: ServiceBindingDraft,
+  fqdn: string,
+  commonFqdns: string[],
+  seenCommon: Set<string>,
+  weights: Record<string, number>,
+  priorities: Record<string, number>,
+): { weights: Record<string, number>; priorities: Record<string, number> } {
+  const key = fqdnKey(fqdn)
+  if (fqdn && key && !seenCommon.has(key)) {
+    seenCommon.add(key)
+    commonFqdns.push(draft.fqdn)
+  }
+  if (Object.keys(weights).length === 0) {
+    return {
+      weights: { ...draft.target_ip_weights },
+      priorities: { ...draft.target_ip_priorities },
+    }
+  }
+  return { weights, priorities }
 }
 
 export function hydrateAddressBlock(
   drafts: ServiceBindingDraft[],
   pool: string[] = [],
 ): AddressBlockState {
-  const primary = drafts[0]
-  const ips = uniqueIps(pool, primary?.target_ips ?? [])
+  const multiIpTargets = drafts
+    .filter((draft) => draft.record_type === 'A' && draft.target_ips.length > 1)
+    .map((draft) => draft.target_ips)
+  const allAIps = drafts
+    .filter((draft) => draft.record_type === 'A')
+    .map((draft) => draft.target_ips)
+  const ips =
+    pool.length > 0
+      ? uniqueIps(pool)
+      : uniqueIps(...(multiIpTargets.length > 0 ? multiIpTargets : allAIps))
   const poolSet = new Set(ips)
-  const claimed = new Set<string>()
-  const extraByIp = new Map<string, string>()
-  const otherBindings: ServiceBindingDraft[] = []
+  const commonFqdns: string[] = []
+  const seenCommon = new Set<string>()
+  const seenExtra = new Set<string>()
+  const extraByIp = new Map<string, string[]>()
+  const preservedBindings: ServiceBindingDraft[] = []
+  let weights: Record<string, number> = {}
+  let priorities: Record<string, number> = {}
+  const splitSinglePool =
+    ips.length === 1 &&
+    drafts.filter((draft) => isFullPoolA(draft, ips) || isCommonPoolA(draft, poolSet))
+      .length > 1
+  let assignedFirstSinglePoolCommon = false
 
-  for (const extra of drafts.slice(1)) {
-    const ip = canCollapseToNode(extra, poolSet, claimed)
-    if (ip) {
-      claimed.add(ip)
-      extraByIp.set(ip, extra.fqdn)
+  function pushExtra(ip: string, fqdn: string) {
+    const key = fqdnKey(fqdn)
+    if (!key || seenExtra.has(key) || seenCommon.has(key)) return
+    seenExtra.add(key)
+    const list = extraByIp.get(ip) ?? []
+    list.push(fqdn)
+    extraByIp.set(ip, list)
+  }
+
+  function promoteToCommon(draft: ServiceBindingDraft, fqdn: string) {
+    const key = fqdnKey(fqdn)
+    if (key && seenExtra.has(key)) {
+      seenExtra.delete(key)
+      for (const [ip, list] of extraByIp) {
+        extraByIp.set(
+          ip,
+          list.filter((item) => fqdnKey(item) !== key),
+        )
+      }
+    }
+    const next = takeAsCommon(
+      draft,
+      fqdn,
+      commonFqdns,
+      seenCommon,
+      weights,
+      priorities,
+    )
+    weights = next.weights
+    priorities = next.priorities
+  }
+
+  for (const draft of drafts) {
+    const fqdn = draft.fqdn.trim()
+    if (splitSinglePool && (isFullPoolA(draft, ips) || isCommonPoolA(draft, poolSet))) {
+      if (!assignedFirstSinglePoolCommon) {
+        assignedFirstSinglePoolCommon = true
+        promoteToCommon(draft, fqdn)
+        continue
+      }
+      const ip = draft.target_ips[0]?.trim() ?? ''
+      if (ip && poolSet.has(ip) && fqdn) {
+        pushExtra(ip, draft.fqdn)
+        continue
+      }
+    }
+    // Full pool OR multi-IP subset of pool → common (heals orphaned toggle damage).
+    if (isFullPoolA(draft, ips) || isCommonPoolA(draft, poolSet)) {
+      promoteToCommon(draft, fqdn)
       continue
     }
-    otherBindings.push(extra)
+    if (draft.record_type === 'A' && draft.target_ips.length === 1) {
+      const ip = draft.target_ips[0]?.trim() ?? ''
+      if (ip && poolSet.has(ip) && fqdn) {
+        pushExtra(ip, draft.fqdn)
+        continue
+      }
+    }
+    preservedBindings.push(draft)
   }
 
   return {
-    commonFqdn: primary?.fqdn ?? '',
+    commonFqdns,
     nodes: ips.map((ip) => ({
       ip,
-      extraFqdn: extraByIp.get(ip) ?? '',
+      extraFqdns: extraByIp.get(ip) ?? [],
     })),
-    otherBindings,
-    target_ip_weights: { ...(primary?.target_ip_weights ?? {}) },
-    target_ip_priorities: { ...(primary?.target_ip_priorities ?? {}) },
+    preservedBindings,
+    target_ip_weights: weights,
+    target_ip_priorities: priorities,
   }
 }
 
@@ -221,7 +304,7 @@ export function removeAddressNode(state: AddressBlockState, ip: string): Address
   return {
     ...state,
     nodes: state.nodes.filter((node) => node.ip !== ip),
-    otherBindings: pruneIpFromBindings(state.otherBindings, ip),
+    preservedBindings: pruneIpFromBindings(state.preservedBindings, ip),
     target_ip_weights: omitKey(state.target_ip_weights, ip),
     target_ip_priorities: omitKey(state.target_ip_priorities, ip),
   }
@@ -234,12 +317,142 @@ export function addAddressNode(state: AddressBlockState, ip: string): AddressBlo
   }
   return {
     ...state,
-    nodes: [...state.nodes, { ip: trimmed, extraFqdn: '' }],
+    nodes: [...state.nodes, { ip: trimmed, extraFqdns: [] }],
     target_ip_weights: { ...state.target_ip_weights, [trimmed]: 1 },
     target_ip_priorities: { ...state.target_ip_priorities, [trimmed]: 1 },
   }
 }
 
+const LB_META_MIN = 1
+const LB_META_MAX = 100
+
+function clampLbMeta(value: number): number {
+  if (!Number.isFinite(value)) return LB_META_MIN
+  return Math.min(LB_META_MAX, Math.max(LB_META_MIN, Math.round(value)))
+}
+
+export function patchAddressIpMeta(
+  state: AddressBlockState,
+  ip: string,
+  meta: { weight?: number; priority?: number },
+): AddressBlockState {
+  if (!state.nodes.some((node) => node.ip === ip)) return state
+  return {
+    ...state,
+    target_ip_weights:
+      meta.weight === undefined
+        ? state.target_ip_weights
+        : { ...state.target_ip_weights, [ip]: clampLbMeta(meta.weight) },
+    target_ip_priorities:
+      meta.priority === undefined
+        ? state.target_ip_priorities
+        : { ...state.target_ip_priorities, [ip]: clampLbMeta(meta.priority) },
+  }
+}
+
+export function addressHasFqdn(state: AddressBlockState, fqdn: string): boolean {
+  const key = fqdnKey(fqdn)
+  if (!key) return false
+  if (state.commonFqdns.some((item) => fqdnKey(item) === key)) return true
+  if (state.nodes.some((node) => node.extraFqdns.some((item) => fqdnKey(item) === key))) {
+    return true
+  }
+  return state.preservedBindings.some((item) => fqdnKey(item.fqdn) === key)
+}
+
+export function addCommonFqdn(state: AddressBlockState, fqdn: string): AddressBlockState {
+  const trimmed = fqdn.trim()
+  if (!trimmed) return state
+  const key = fqdnKey(trimmed)
+  if (state.commonFqdns.some((item) => fqdnKey(item) === key)) return state
+  if (state.nodes.some((node) => node.extraFqdns.some((item) => fqdnKey(item) === key))) {
+    return state
+  }
+  // Promote out of invisible preserved (corrupted / CNAME-adjacent duplicates).
+  const preservedBindings = state.preservedBindings.filter(
+    (item) => fqdnKey(item.fqdn) !== key,
+  )
+  return {
+    ...state,
+    commonFqdns: [...state.commonFqdns, trimmed],
+    preservedBindings,
+  }
+}
+
+export function removeCommonFqdn(state: AddressBlockState, index: number): AddressBlockState {
+  return {
+    ...state,
+    commonFqdns: state.commonFqdns.filter((_, i) => i !== index),
+  }
+}
+
+export function updateCommonFqdn(
+  state: AddressBlockState,
+  index: number,
+  fqdn: string,
+): AddressBlockState {
+  return {
+    ...state,
+    commonFqdns: state.commonFqdns.map((item, i) => (i === index ? fqdn : item)),
+  }
+}
+
+export function addExtraFqdn(
+  state: AddressBlockState,
+  ip: string,
+  fqdn: string,
+): AddressBlockState {
+  const trimmed = fqdn.trim()
+  if (!trimmed) return state
+  const key = fqdnKey(trimmed)
+  if (state.commonFqdns.some((item) => fqdnKey(item) === key)) return state
+  if (state.nodes.some((node) => node.extraFqdns.some((item) => fqdnKey(item) === key))) {
+    return state
+  }
+  if (!state.nodes.some((node) => node.ip === ip)) return state
+  return {
+    ...state,
+    preservedBindings: state.preservedBindings.filter((item) => fqdnKey(item.fqdn) !== key),
+    nodes: state.nodes.map((node) =>
+      node.ip === ip ? { ...node, extraFqdns: [...node.extraFqdns, trimmed] } : node,
+    ),
+  }
+}
+
+export function removeExtraFqdn(
+  state: AddressBlockState,
+  ip: string,
+  index: number,
+): AddressBlockState {
+  return {
+    ...state,
+    nodes: state.nodes.map((node) =>
+      node.ip === ip
+        ? { ...node, extraFqdns: node.extraFqdns.filter((_, i) => i !== index) }
+        : node,
+    ),
+  }
+}
+
+export function updateExtraFqdn(
+  state: AddressBlockState,
+  ip: string,
+  index: number,
+  fqdn: string,
+): AddressBlockState {
+  return {
+    ...state,
+    nodes: state.nodes.map((node) =>
+      node.ip === ip
+        ? {
+            ...node,
+            extraFqdns: node.extraFqdns.map((item, i) => (i === index ? fqdn : item)),
+          }
+        : node,
+    ),
+  }
+}
+
 export function toAddressBindings(
   state: AddressBlockState,
   primary: AddressPrimaryMeta,
@@ -253,10 +466,11 @@ export function toAddressBindings(
   )
 
   const drafts: ServiceBindingDraft[] = []
-  const hasPrimary = Boolean(state.commonFqdn.trim()) || ips.length > 0
-  if (hasPrimary) {
+  for (const raw of state.commonFqdns) {
+    const fqdn = raw.trim()
+    if (!fqdn || ips.length === 0) continue
     drafts.push({
-      fqdn: state.commonFqdn,
+      fqdn,
       record_type: 'A',
       target_ips: ips,
       target_cname: '',
@@ -268,21 +482,29 @@ export function toAddressBindings(
   }
 
   for (const node of state.nodes) {
-    const extraFqdn = node.extraFqdn.trim()
-    if (!extraFqdn) continue
-    drafts.push({
-      fqdn: extraFqdn,
-      record_type: 'A',
-      target_ips: [node.ip],
-      target_cname: '',
-      lb_mode: primary.lb_mode,
-      health: { ...primary.health },
-      target_ip_weights: { [node.ip]: weights[node.ip] ?? 1 },
-      target_ip_priorities: { [node.ip]: priorities[node.ip] ?? 1 },
-    })
+    for (const raw of node.extraFqdns) {
+      const extraFqdn = raw.trim()
+      if (!extraFqdn) continue
+      drafts.push({
+        fqdn: extraFqdn,
+        record_type: 'A',
+        target_ips: [node.ip],
+        target_cname: '',
+        lb_mode: primary.lb_mode,
+        health: { ...primary.health },
+        target_ip_weights: { [node.ip]: weights[node.ip] ?? 1 },
+        target_ip_priorities: { [node.ip]: priorities[node.ip] ?? 1 },
+      })
+    }
   }
 
-  drafts.push(...state.otherBindings)
+  const seen = new Set(drafts.map((item) => fqdnKey(item.fqdn)).filter(Boolean))
+  for (const preserved of state.preservedBindings) {
+    const key = fqdnKey(preserved.fqdn)
+    if (!key || seen.has(key)) continue
+    seen.add(key)
+    drafts.push(preserved)
+  }
   return drafts
 }
 

apps/web/src/queries/services.ts

@@ -1,6 +1,7 @@
 import { queryOptions } from '@tanstack/react-query'
 import { api } from '@/lib/api-client'
 import {
+  failoverLogSchema,
   healthProbeLogSchema,
   serviceBindingSchema,
   serviceGroupsResponseSchema,
@@ -35,6 +36,8 @@ export const serviceGroupsQueryOptions = () =>
       }
       return parsed.data
     },
+    refetchInterval: 10_000,
+    staleTime: 5_000,
   })
 
 export const servicesQueryOptions = () =>
@@ -89,6 +92,7 @@ export const serviceDetailKeys = {
   overview: (id: number) => [...serviceKeys.all, id, 'overview'] as const,
   nodes: (id: number) => [...serviceKeys.all, id, 'nodes'] as const,
   healthLog: (id: number) => [...serviceKeys.all, id, 'health-log'] as const,
+  failoverLog: (id: number) => [...serviceKeys.all, id, 'failover-log'] as const,
   view: (id: number) => [...serviceKeys.all, id, 'view'] as const,
 }
 
@@ -99,6 +103,8 @@ export const serviceViewQueryOptions = (id: number) =>
       const data = await api.get<unknown>(`/api/v1/services/${id}`)
       return serviceViewSchema.parse(data)
     },
+    refetchInterval: 10_000,
+    staleTime: 5_000,
   })
 
 export const serviceHealthLogQueryOptions = (id: number) =>
@@ -108,6 +114,17 @@ export const serviceHealthLogQueryOptions = (id: number) =>
       const data = await api.get<unknown>(`/api/v1/services/${id}/health-log`)
       return z.object({ items: z.array(healthProbeLogSchema) }).parse(data)
     },
+    refetchInterval: 10_000,
+    staleTime: 5_000,
+  })
+
+export const serviceFailoverLogQueryOptions = (id: number) =>
+  queryOptions({
+    queryKey: serviceDetailKeys.failoverLog(id),
+    queryFn: async () => {
+      const data = await api.get<unknown>(`/api/v1/services/${id}/failover-log`)
+      return z.object({ items: z.array(failoverLogSchema) }).parse(data)
+    },
   })
 
 export const serviceOverviewQueryOptions = (id: number) =>

apps/web/src/routes/_auth/services/$serviceId/index.tsx

@@ -14,7 +14,6 @@ import {
 import { ChangeDomainSheet } from '@/components/change-domain-sheet'
 import { ChangeIpSheet } from '@/components/change-ip-sheet'
 import { EmptyState } from '@/components/empty-state'
-import { FailoverTimeline } from '@/components/failover-timeline'
 import { FormFieldSimple } from '@/components/form-field'
 import { FormSheet } from '@/components/form-sheet'
 import { HealthCheckBadge } from '@/components/health-check-badge'
@@ -29,19 +28,15 @@ import {
 import { LbModeTile } from '@/components/services/service-unit-card'
 import {
   KpiStatGrid,
+  ServiceFailoverPanel,
   ServiceHealthMonitor,
 } from '@/components/reui-kit'
-import {
-  Frame,
-  FrameDescription,
-  FrameHeader,
-  FramePanel,
-  FrameTitle,
-} from '@/components/reui/frame'
 import { api } from '@/lib/api-client'
+import { hasSharedPool, uniqueIpCount } from '@/lib/failover-events'
 import {
   enabledHealthProviders,
   providerHealthStatuses,
+  resolveServiceDisplayHealth,
 } from '@/lib/health-log'
 import type { ServiceView, UpdateServiceConfigInput } from '@/lib/schemas'
 import {
@@ -51,6 +46,7 @@ import {
   domainsListQueryOptions,
   serviceBindingKeys,
   serviceDetailKeys,
+  serviceFailoverLogQueryOptions,
   serviceGroupKeys,
   serviceGroupsQueryOptions,
   serviceHealthLogQueryOptions,
@@ -84,6 +80,7 @@ interface OverviewPayload {
     priority: number
     consecutive_failures: number
     last_failure_reason: string | null
+    last_check_at?: string | null
   }>
 }
 
@@ -96,6 +93,7 @@ function ServiceDetailPage() {
   const viewQuery = useQuery(serviceViewQueryOptions(id))
   const overviewQuery = useQuery(serviceOverviewQueryOptions(id))
   const logQuery = useQuery(serviceHealthLogQueryOptions(id))
+  const failoverLogQuery = useQuery(serviceFailoverLogQueryOptions(id))
   const nodesQuery = useQuery(serviceNodesQueryOptions(id))
   const groupsQuery = useQuery(serviceGroupsQueryOptions())
   const domainsQuery = useQuery(domainsListQueryOptions())
@@ -106,6 +104,35 @@ function ServiceDetailPage() {
     () => logQuery.data?.items ?? [],
     [logQuery.data?.items],
   )
+  const failoverHistory = useMemo(
+    () => failoverLogQuery.data?.items ?? [],
+    [failoverLogQuery.data?.items],
+  )
+  const failoverBindings = useMemo(
+    () =>
+      (service?.domains ?? []).map((domain) => ({
+        fqdn: domain.fqdn,
+        configured: domain.target_ips,
+        active: domain.active_ips ?? [],
+      })),
+    [service?.domains],
+  )
+  const uniqueEnabledIps = useMemo(
+    () =>
+      (service?.ips ?? []).filter((ip) => service?.ip_enabled[ip] !== false),
+    [service?.ips, service?.ip_enabled],
+  )
+  const displayHealth = useMemo(
+    () =>
+      resolveServiceDisplayHealth(
+        service?.health_status,
+        service?.ip_health ?? [],
+        logItems,
+      ),
+    [service?.health_status, service?.ip_health, logItems],
+  )
+  const showPoolPanel =
+    uniqueEnabledIps.length >= 2 && hasSharedPool(failoverBindings)
   const nodes = (nodesQuery.data as OverviewPayload['nodes']) ?? []
 
   const [editOpen, setEditOpen] = useState(false)
@@ -135,6 +162,7 @@ function ServiceDetailPage() {
       queryClient.invalidateQueries({ queryKey: serviceDetailKeys.overview(id) }),
       queryClient.invalidateQueries({ queryKey: serviceDetailKeys.nodes(id) }),
       queryClient.invalidateQueries({ queryKey: serviceDetailKeys.healthLog(id) }),
+      queryClient.invalidateQueries({ queryKey: serviceDetailKeys.failoverLog(id) }),
     ])
   }
 
@@ -209,22 +237,6 @@ function ServiceDetailPage() {
   const isError = viewQuery.isError || overviewQuery.isError
   const error = viewQuery.error ?? overviewQuery.error
 
-  const failoverEvents =
-    (nodes.length > 0 ? nodes : (overview?.nodes ?? []))
-      .filter(
-        (node) =>
-          node.health_status === 'unhealthy' ||
-          node.health_status === 'down' ||
-          node.health_status === 'checking',
-      )
-      .map((node) => ({
-        id: node.address,
-        title: `${node.address}: ${node.health_status}`,
-        detail: node.last_failure_reason
-          ? `${node.last_failure_reason} · fail ${node.consecutive_failures}`
-          : `fail ${node.consecutive_failures}`,
-      }))
-
   const enabledProviders = useMemo(
     () => enabledHealthProviders(service?.domains ?? []),
     [service],
@@ -256,8 +268,11 @@ function ServiceDetailPage() {
             description="Domain → Service → Node → Health → Failover"
             actions={
               <>
-                <LbModeTile mode={service.lb_mode} />
-                <HealthCheckBadge status={service.health_status} />
+                <LbModeTile
+                  mode={service.lb_mode}
+                  hasPool={uniqueIpCount(service.ips) >= 2}
+                />
+                <HealthCheckBadge status={displayHealth} />
                 <Tooltip>
                   <TooltipTrigger
                     render={
@@ -284,20 +299,20 @@ function ServiceDetailPage() {
                 id: 'status',
                 icon: <ActivityIcon />,
                 label: 'Статус',
-                value: service.health_status === 'up' ? 'OK' : service.health_status,
+                value: displayHealth === 'up' ? 'OK' : displayHealth,
                 variant:
-                  service.health_status === 'down'
+                  displayHealth === 'down'
                     ? 'destructive'
-                    : service.health_status === 'degraded'
+                    : displayHealth === 'degraded'
                       ? 'warning'
                       : 'default',
                 iconClassName:
-                  service.health_status === 'down'
+                  displayHealth === 'down'
                     ? 'text-destructive'
-                    : service.health_status === 'degraded'
+                    : displayHealth === 'degraded'
                       ? 'text-warning'
                       : 'text-success',
-                hint: <HealthCheckBadge status={service.health_status} size="xs" />,
+                hint: <HealthCheckBadge status={displayHealth} size="xs" />,
               },
               {
                 id: 'fqdn',
@@ -311,21 +326,33 @@ function ServiceDetailPage() {
                 icon: <NetworkIcon />,
                 label: 'IP',
                 value: String(service.ips.length),
-                hint: `${service.active_ips.length} в пуле`,
+                hint: showPoolPanel
+                  ? `${service.active_ips.length} в пуле`
+                  : uniqueEnabledIps.length <= 1
+                    ? 'без балансировки'
+                    : service.ips.join(', ') || 'нет',
               },
               {
                 id: 'pool',
                 icon: <ServerIcon />,
-                label: 'Активный пул',
-                value: String((overview?.active_addresses ?? service.active_ips).length),
-                hint: (overview?.active_addresses ?? service.active_ips).join(', ') || 'нет',
+                label: showPoolPanel ? 'Активный пул' : 'Адреса',
+                value: String(
+                  (overview?.active_addresses ?? service.active_ips).length,
+                ),
+                hint:
+                  (overview?.active_addresses ?? service.active_ips).join(', ') ||
+                  'нет',
               },
             ]}
           />
 
           <section
             aria-label="Мониторинг"
-            className="grid min-w-0 items-start gap-2 @3xl:grid-cols-2"
+            className={
+              showPoolPanel
+                ? 'grid min-w-0 items-start gap-2 @3xl:grid-cols-2'
+                : 'grid min-w-0 items-start gap-2'
+            }
           >
             <ServiceHealthMonitor
               items={logItems}
@@ -333,17 +360,16 @@ function ServiceDetailPage() {
               statuses={providerStatuses}
               isLoading={logQuery.isLoading}
             />
-            <Frame dense spacing="sm" className="min-w-0 w-full">
-              <FrameHeader>
-                <FrameTitle>Failover</FrameTitle>
-                <FrameDescription>
-                  Нездоровые ноды и причины последней ошибки
-                </FrameDescription>
-              </FrameHeader>
-              <FramePanel>
-                <FailoverTimeline events={failoverEvents} />
-              </FramePanel>
-            </Frame>
+            {showPoolPanel ? (
+              <ServiceFailoverPanel
+                lbMode={service.lb_mode}
+                hasPool={uniqueIpCount(service.ips) >= 2}
+                ipHealth={service.ip_health}
+                bindings={failoverBindings}
+                history={failoverHistory}
+                probes={logItems}
+              />
+            ) : null}
           </section>
 
           {service.ips.length === 0 && service.domains.length === 0 ? (
@@ -356,6 +382,7 @@ function ServiceDetailPage() {
             <ServiceDetailGrid
               service={service}
               nodes={nodes}
+              probes={logItems}
               togglingIp={togglingIp}
               onToggleIp={(ip, enabled) => {
                 setTogglingIp(ip)

apps/web/src/routes/_auth/services/$serviceId/route.tsx

@@ -1,6 +1,7 @@
 import { createFileRoute, Outlet } from '@tanstack/react-router'
 import { PageShell } from '@/components/page-shell'
 import {
+  serviceFailoverLogQueryOptions,
   serviceHealthLogQueryOptions,
   serviceNodesQueryOptions,
   serviceOverviewQueryOptions,
@@ -14,6 +15,7 @@ export const Route = createFileRoute('/_auth/services/$serviceId')({
       queryClient.ensureQueryData(serviceViewQueryOptions(id)),
       queryClient.ensureQueryData(serviceOverviewQueryOptions(id)),
       queryClient.ensureQueryData(serviceHealthLogQueryOptions(id)),
+      queryClient.ensureQueryData(serviceFailoverLogQueryOptions(id)),
       queryClient.ensureQueryData(serviceNodesQueryOptions(id)),
     ])
     return { breadcrumb: view.name }

docs/Home.md

@@ -44,9 +44,10 @@ health-check работают на двух уровнях:
 - **Change Domain** — перенос привязок между зонами `POST /api/v1/services/:id/change-domain`.
 
 Режимы LB: `round_robin`, `failover`, `weighted`. В Cloudflare free `weighted`
-работает как `round_robin` (одна A на IP). `unknown` **не** считается healthy и
-не попадает в пул, пока нет успешных проб; восстановление — `UNHEALTHY → CHECKING → HEALTHY`
-после `HEALTH_SUCCESS_RECOVERIES` (default 2). Пороги и cron движка задаются в
+работает как `round_robin` (одна A на IP). В A-пул попадает всё, кроме **Down**
+(`unknown` / checking / Slow возвращаются в DNS на первой успешной пробе).
+Бейдж Healthy — `UNHEALTHY → CHECKING → HEALTHY` после `HEALTH_SUCCESS_RECOVERIES`
+(default 2). Пороги и cron движка задаются в
 **Настройки → Health-check** (env — fallback, пока значения не сохранены в UI).
 
 ### Источники проб: Local, Cloudflare Worker, Globalping

packages/db/dist/index.js

@@ -355,6 +355,15 @@ var notificationLog = sqliteTable("notification_log", {
   message: text("message").notNull(),
   created_at: text("created_at").notNull().default(sql`datetime('now')`)
 });
+var failoverLog = sqliteTable("failover_log", {
+  id: integer("id").primaryKey({ autoIncrement: true }),
+  service_id: integer("service_id").notNull().references(() => services.id, { onDelete: "cascade" }),
+  binding_id: integer("binding_id").notNull().references(() => serviceBindings.id, { onDelete: "cascade" }),
+  fqdn: text("fqdn").notNull(),
+  ip: text("ip").notNull(),
+  action: text("action").notNull(),
+  created_at: text("created_at").notNull().default(sql`datetime('now')`)
+});
 var auditLog = sqliteTable("audit_log", {
   id: text("id").primaryKey(),
   event_id: text("event_id"),
@@ -395,6 +404,7 @@ var schema = {
   domainMonitorResults,
   healthProbeLog,
   notificationLog,
+  failoverLog,
   auditLog
 };
 
@@ -715,6 +725,7 @@ __export(repos_exports, {
   getSyncJob: () => getSyncJob,
   insertBinding: () => insertBinding,
   insertDnsRecord: () => insertDnsRecord,
+  insertFailoverLog: () => insertFailoverLog,
   insertHealthProbeLog: () => insertHealthProbeLog,
   insertNotificationLog: () => insertNotificationLog,
   linkBindingRecord: () => linkBindingRecord,
@@ -738,6 +749,7 @@ __export(repos_exports, {
   listDomains: () => listDomains,
   listDomainsEnriched: () => listDomainsEnriched,
   listEnabledDomainMonitors: () => listEnabledDomainMonitors,
+  listFailoverLogForService: () => listFailoverLogForService,
   listGroupDnsRecords: () => listGroupDnsRecords,
   listGroups: () => listGroups,
   listHealthCheckTargets: () => listHealthCheckTargets,
@@ -1995,6 +2007,12 @@ var WORST_HEALTH_SQL = sql2.raw(`CASE
   END) = 1 THEN 'up'
   ELSE 'unknown'
 END`);
+var BEST_ALIVE_HEALTH_SQL = sql2.raw(`CASE
+  WHEN MAX(CASE WHEN status = 'up' THEN 1 ELSE 0 END) = 1 THEN 'up'
+  WHEN MAX(CASE WHEN status = 'degraded' THEN 1 ELSE 0 END) = 1 THEN 'degraded'
+  WHEN MAX(CASE WHEN status = 'down' THEN 1 ELSE 0 END) = 1 THEN 'down'
+  ELSE 'unknown'
+END`);
 function aggregateIpHealthByRefs(db, scope, refIds) {
   const result = /* @__PURE__ */ new Map();
   if (refIds.length === 0) return result;
@@ -2055,7 +2073,7 @@ function aggregateIpHealthByServiceIds(db, serviceIds) {
   );
   const rows = db.all(sql2`
     SELECT sb.service_id AS service_id,
-      ${WORST_HEALTH_SQL} AS health_status,
+      ${BEST_ALIVE_HEALTH_SQL} AS health_status,
       MAX(ihs.latency_ms) AS health_latency_ms
     FROM ip_health_status ihs
     INNER JOIN service_bindings sb
@@ -2545,6 +2563,26 @@ function listNotificationLog(db, limit = 50) {
     LIMIT ${limit}
   `);
 }
+function insertFailoverLog(db, input) {
+  for (const entry of input.entries) {
+    db.insert(failoverLog).values({
+      service_id: input.serviceId,
+      binding_id: input.bindingId,
+      fqdn: input.fqdn,
+      ip: entry.ip,
+      action: entry.action
+    }).run();
+  }
+}
+function listFailoverLogForService(db, serviceId, limit = 100) {
+  return db.all(sql2`
+    SELECT id, service_id, binding_id, fqdn, ip, action, created_at
+    FROM failover_log
+    WHERE service_id = ${serviceId}
+    ORDER BY created_at DESC, id DESC
+    LIMIT ${limit}
+  `);
+}
 export {
   ConflictError,
   NotFoundError,
@@ -2560,6 +2598,7 @@ export {
   domainMonitors,
   domainTags,
   domains,
+  failoverLog,
   getAppSettings,
   getAppSettingsSecrets,
   groups,

packages/db/migrations/025_failover_log.sql

@@ -0,0 +1,11 @@
+CREATE TABLE failover_log (
+  id INTEGER PRIMARY KEY AUTOINCREMENT,
+  service_id INTEGER NOT NULL REFERENCES services(id) ON DELETE CASCADE,
+  binding_id INTEGER NOT NULL REFERENCES service_bindings(id) ON DELETE CASCADE,
+  fqdn TEXT NOT NULL,
+  ip TEXT NOT NULL,
+  action TEXT NOT NULL CHECK (action IN ('added', 'removed')),
+  created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+
+CREATE INDEX idx_failover_log_service ON failover_log(service_id, created_at DESC);

packages/db/src/repos.ts

@@ -49,6 +49,7 @@ import {
   nodes,
   bindingNodes,
   notificationLog,
+  failoverLog,
   serviceBindingIps,
   serviceBindingRecords,
   serviceBindings,
@@ -2163,6 +2164,14 @@ const WORST_HEALTH_SQL = sql.raw(`CASE
   ELSE 'unknown'
 END`);
 
+/** Service is up if any binding IP is up. */
+const BEST_ALIVE_HEALTH_SQL = sql.raw(`CASE
+  WHEN MAX(CASE WHEN status = 'up' THEN 1 ELSE 0 END) = 1 THEN 'up'
+  WHEN MAX(CASE WHEN status = 'degraded' THEN 1 ELSE 0 END) = 1 THEN 'degraded'
+  WHEN MAX(CASE WHEN status = 'down' THEN 1 ELSE 0 END) = 1 THEN 'down'
+  ELSE 'unknown'
+END`);
+
 /** Worst status across ip_health_status rows for each ref_id in a scope. */
 export function aggregateIpHealthByRefs(
   db: Db,
@@ -2236,7 +2245,7 @@ export function aggregateGroupScopeHealthByIds(
   return result;
 }
 
-/** Worst binding-scope health rolled up per service_id. */
+/** Binding-scope health rolled up per service_id: any up → service up. */
 export function aggregateIpHealthByServiceIds(
   db: Db,
   serviceIds: number[],
@@ -2253,7 +2262,7 @@ export function aggregateIpHealthByServiceIds(
     health_latency_ms: number | null;
   }>(sql`
     SELECT sb.service_id AS service_id,
-      ${WORST_HEALTH_SQL} AS health_status,
+      ${BEST_ALIVE_HEALTH_SQL} AS health_status,
       MAX(ihs.latency_ms) AS health_latency_ms
     FROM ip_health_status ihs
     INNER JOIN service_bindings sb
@@ -2332,6 +2341,113 @@ export function listIpHealthByServiceIds(
   return result;
 }
 
+function parseIpHealthState(status: string | null): IpHealthState | undefined {
+  if (
+    status === "up" ||
+    status === "down" ||
+    status === "degraded" ||
+    status === "unknown"
+  ) {
+    return status;
+  }
+  return undefined;
+}
+
+function bestAliveIpState(statuses: readonly IpHealthState[]): IpHealthState {
+  if (statuses.some((status) => status === "up")) return "up";
+  if (statuses.some((status) => status === "degraded")) return "degraded";
+  if (statuses.some((status) => status === "down")) return "down";
+  return "unknown";
+}
+
+/**
+ * Latest probe per service+IP+provider from health_probe_log, then any-up per IP.
+ * Display overlay — hysteresis in ip_health_status is unchanged (DNS).
+ */
+export function listLatestLiveHealthByServiceIds(
+  db: Db,
+  serviceIds: number[],
+): Map<number, ServiceIpHealthRow[]> {
+  const result = new Map<number, ServiceIpHealthRow[]>();
+  if (serviceIds.length === 0) return result;
+  const idList = sql.join(
+    serviceIds.map((id) => sql`${id}`),
+    sql`, `,
+  );
+  const rows = db.all<{
+    service_id: number;
+    ip: string;
+    provider: string | null;
+    status: string | null;
+    latency_ms: number | null;
+    last_error: string | null;
+    colo: string | null;
+    checked_at: string | null;
+  }>(sql`
+    SELECT sb.service_id AS service_id,
+      l.ip AS ip,
+      l.provider AS provider,
+      l.status AS status,
+      l.latency_ms AS latency_ms,
+      l.error AS last_error,
+      l.colo AS colo,
+      l.checked_at AS checked_at
+    FROM health_probe_log l
+    INNER JOIN service_bindings sb
+      ON l.scope = 'binding' AND l.ref_id = sb.id
+    INNER JOIN (
+      SELECT sb2.service_id AS service_id,
+        l2.ip AS ip,
+        l2.provider AS provider,
+        MAX(l2.id) AS max_id
+      FROM health_probe_log l2
+      INNER JOIN service_bindings sb2
+        ON l2.scope = 'binding' AND l2.ref_id = sb2.id
+      WHERE sb2.service_id IN (${idList})
+      GROUP BY sb2.service_id, l2.ip, l2.provider
+    ) latest
+      ON latest.max_id = l.id
+  `);
+
+  const byServiceIp = new Map<
+    string,
+    { serviceId: number; ip: string; probes: ServiceIpHealthRow[] }
+  >();
+  for (const row of rows) {
+    const status = parseIpHealthState(row.status);
+    if (!status) continue;
+    const key = `${row.service_id}\0${row.ip}`;
+    const probe: ServiceIpHealthRow = {
+      ip: row.ip,
+      status,
+      latency_ms: row.latency_ms,
+      last_checked_at: row.checked_at,
+      last_error: row.last_error,
+      provider: normalizeStatusProvider(row.provider),
+      colo: row.colo,
+    };
+    const bucket = byServiceIp.get(key);
+    if (bucket) bucket.probes.push(probe);
+    else {
+      byServiceIp.set(key, {
+        serviceId: row.service_id,
+        ip: row.ip,
+        probes: [probe],
+      });
+    }
+  }
+
+  for (const { serviceId, ip, probes } of byServiceIp.values()) {
+    const status = bestAliveIpState(probes.map((probe) => probe.status));
+    const preferred =
+      probes.find((probe) => probe.status === status) ?? probes[0]!;
+    const list = result.get(serviceId) ?? [];
+    list.push({ ...preferred, ip, status });
+    result.set(serviceId, list);
+  }
+  return result;
+}
+
 export function mergeHealthAggregates(
   parts: Array<HealthAggregate | undefined | null>,
 ): HealthAggregate {
@@ -3055,3 +3171,51 @@ export function listNotificationLog(
   `);
 }
 
+export type FailoverLogAction = "added" | "removed";
+
+export type FailoverLogRow = {
+  id: number;
+  service_id: number;
+  binding_id: number;
+  fqdn: string;
+  ip: string;
+  action: FailoverLogAction;
+  created_at: string;
+};
+
+export function insertFailoverLog(
+  db: Db,
+  input: {
+    serviceId: number;
+    bindingId: number;
+    fqdn: string;
+    entries: Array<{ ip: string; action: FailoverLogAction }>;
+  },
+): void {
+  for (const entry of input.entries) {
+    db.insert(failoverLog)
+      .values({
+        service_id: input.serviceId,
+        binding_id: input.bindingId,
+        fqdn: input.fqdn,
+        ip: entry.ip,
+        action: entry.action,
+      })
+      .run();
+  }
+}
+
+export function listFailoverLogForService(
+  db: Db,
+  serviceId: number,
+  limit = 100,
+): FailoverLogRow[] {
+  return db.all<FailoverLogRow>(sql`
+    SELECT id, service_id, binding_id, fqdn, ip, action, created_at
+    FROM failover_log
+    WHERE service_id = ${serviceId}
+    ORDER BY created_at DESC, id DESC
+    LIMIT ${limit}
+  `);
+}
+

packages/db/src/schema.ts

@@ -497,6 +497,22 @@ export const notificationLog = sqliteTable("notification_log", {
     .default(sql`datetime('now')`),
 });
 
+export const failoverLog = sqliteTable("failover_log", {
+  id: integer("id").primaryKey({ autoIncrement: true }),
+  service_id: integer("service_id")
+    .notNull()
+    .references(() => services.id, { onDelete: "cascade" }),
+  binding_id: integer("binding_id")
+    .notNull()
+    .references(() => serviceBindings.id, { onDelete: "cascade" }),
+  fqdn: text("fqdn").notNull(),
+  ip: text("ip").notNull(),
+  action: text("action").notNull(),
+  created_at: text("created_at")
+    .notNull()
+    .default(sql`datetime('now')`),
+});
+
 export const auditLog = sqliteTable("audit_log", {
   id: text("id").primaryKey(),
   event_id: text("event_id"),
@@ -540,5 +556,6 @@ export const schema = {
   domainMonitorResults,
   healthProbeLog,
   notificationLog,
+  failoverLog,
   auditLog,
 };

packages/shared/dist/index.d.ts

@@ -143,6 +143,7 @@ interface ServiceDomainBindingView {
     fqdn: string;
     record_type: "A" | "CNAME";
     target_ips: string[];
+    active_ips: string[];
     target_ip_weights: Record<string, number>;
     target_ip_priorities: Record<string, number>;
     target_cname: string | null;
@@ -505,6 +506,19 @@ declare const healthProbeLogSchema: z.ZodObject<{
     checked_at: z.ZodString;
 }, z.core.$strip>;
 type HealthProbeLog = z.infer<typeof healthProbeLogSchema>;
+declare const failoverLogSchema: z.ZodObject<{
+    id: z.ZodNumber;
+    service_id: z.ZodNumber;
+    binding_id: z.ZodNumber;
+    fqdn: z.ZodString;
+    ip: z.ZodString;
+    action: z.ZodEnum<{
+        added: "added";
+        removed: "removed";
+    }>;
+    created_at: z.ZodString;
+}, z.core.$strip>;
+type FailoverLogEntry = z.infer<typeof failoverLogSchema>;
 declare const groupSchema: z.ZodObject<{
     id: z.ZodNumber;
     name: z.ZodString;
@@ -643,6 +657,7 @@ declare const serviceDomainBindingSchema: z.ZodPipe<z.ZodObject<{
         skipped: "skipped";
     }>>;
     sync_status: z.ZodDefault<z.ZodNullable<z.ZodString>>;
+    active_ips: z.ZodDefault<z.ZodArray<z.ZodString>>;
 }, z.core.$strip>, z.ZodTransform<{
     target_ips: string[];
     target_ip_weights: Record<string, number>;
@@ -668,6 +683,7 @@ declare const serviceDomainBindingSchema: z.ZodPipe<z.ZodObject<{
     health_check_aggregate: "any" | "all" | "majority";
     cert_monitoring: "auto" | "required" | "skipped";
     sync_status: string | null;
+    active_ips: string[];
     target_ip?: string | null | undefined;
 }, {
     binding_id: number;
@@ -690,6 +706,7 @@ declare const serviceDomainBindingSchema: z.ZodPipe<z.ZodObject<{
     health_check_aggregate: "any" | "all" | "majority";
     cert_monitoring: "auto" | "required" | "skipped";
     sync_status: string | null;
+    active_ips: string[];
     target_ips?: string[] | undefined;
     target_ip?: string | null | undefined;
     target_ip_weights?: Record<string, number> | undefined;
@@ -763,6 +780,7 @@ declare const serviceViewSchema: z.ZodObject<{
             skipped: "skipped";
         }>>;
         sync_status: z.ZodDefault<z.ZodNullable<z.ZodString>>;
+        active_ips: z.ZodDefault<z.ZodArray<z.ZodString>>;
     }, z.core.$strip>, z.ZodTransform<{
         target_ips: string[];
         target_ip_weights: Record<string, number>;
@@ -788,6 +806,7 @@ declare const serviceViewSchema: z.ZodObject<{
         health_check_aggregate: "any" | "all" | "majority";
         cert_monitoring: "auto" | "required" | "skipped";
         sync_status: string | null;
+        active_ips: string[];
         target_ip?: string | null | undefined;
     }, {
         binding_id: number;
@@ -810,6 +829,7 @@ declare const serviceViewSchema: z.ZodObject<{
         health_check_aggregate: "any" | "all" | "majority";
         cert_monitoring: "auto" | "required" | "skipped";
         sync_status: string | null;
+        active_ips: string[];
         target_ips?: string[] | undefined;
         target_ip?: string | null | undefined;
         target_ip_weights?: Record<string, number> | undefined;
@@ -965,6 +985,7 @@ declare const serviceGroupViewSchema: z.ZodObject<{
                 skipped: "skipped";
             }>>;
             sync_status: z.ZodDefault<z.ZodNullable<z.ZodString>>;
+            active_ips: z.ZodDefault<z.ZodArray<z.ZodString>>;
         }, z.core.$strip>, z.ZodTransform<{
             target_ips: string[];
             target_ip_weights: Record<string, number>;
@@ -990,6 +1011,7 @@ declare const serviceGroupViewSchema: z.ZodObject<{
             health_check_aggregate: "any" | "all" | "majority";
             cert_monitoring: "auto" | "required" | "skipped";
             sync_status: string | null;
+            active_ips: string[];
             target_ip?: string | null | undefined;
         }, {
             binding_id: number;
@@ -1012,6 +1034,7 @@ declare const serviceGroupViewSchema: z.ZodObject<{
             health_check_aggregate: "any" | "all" | "majority";
             cert_monitoring: "auto" | "required" | "skipped";
             sync_status: string | null;
+            active_ips: string[];
             target_ips?: string[] | undefined;
             target_ip?: string | null | undefined;
             target_ip_weights?: Record<string, number> | undefined;
@@ -1176,6 +1199,7 @@ declare const serviceGroupsResponseSchema: z.ZodObject<{
                     skipped: "skipped";
                 }>>;
                 sync_status: z.ZodDefault<z.ZodNullable<z.ZodString>>;
+                active_ips: z.ZodDefault<z.ZodArray<z.ZodString>>;
             }, z.core.$strip>, z.ZodTransform<{
                 target_ips: string[];
                 target_ip_weights: Record<string, number>;
@@ -1201,6 +1225,7 @@ declare const serviceGroupsResponseSchema: z.ZodObject<{
                 health_check_aggregate: "any" | "all" | "majority";
                 cert_monitoring: "auto" | "required" | "skipped";
                 sync_status: string | null;
+                active_ips: string[];
                 target_ip?: string | null | undefined;
             }, {
                 binding_id: number;
@@ -1223,6 +1248,7 @@ declare const serviceGroupsResponseSchema: z.ZodObject<{
                 health_check_aggregate: "any" | "all" | "majority";
                 cert_monitoring: "auto" | "required" | "skipped";
                 sync_status: string | null;
+                active_ips: string[];
                 target_ips?: string[] | undefined;
                 target_ip?: string | null | undefined;
                 target_ip_weights?: Record<string, number> | undefined;
@@ -1338,6 +1364,7 @@ declare const serviceGroupsResponseSchema: z.ZodObject<{
                 skipped: "skipped";
             }>>;
             sync_status: z.ZodDefault<z.ZodNullable<z.ZodString>>;
+            active_ips: z.ZodDefault<z.ZodArray<z.ZodString>>;
         }, z.core.$strip>, z.ZodTransform<{
             target_ips: string[];
             target_ip_weights: Record<string, number>;
@@ -1363,6 +1390,7 @@ declare const serviceGroupsResponseSchema: z.ZodObject<{
             health_check_aggregate: "any" | "all" | "majority";
             cert_monitoring: "auto" | "required" | "skipped";
             sync_status: string | null;
+            active_ips: string[];
             target_ip?: string | null | undefined;
         }, {
             binding_id: number;
@@ -1385,6 +1413,7 @@ declare const serviceGroupsResponseSchema: z.ZodObject<{
             health_check_aggregate: "any" | "all" | "majority";
             cert_monitoring: "auto" | "required" | "skipped";
             sync_status: string | null;
+            active_ips: string[];
             target_ips?: string[] | undefined;
             target_ip?: string | null | undefined;
             target_ip_weights?: Record<string, number> | undefined;
@@ -2218,6 +2247,11 @@ declare const cfdmBindingSyncItemSchema: z.ZodObject<{
     hostname: z.ZodString;
     ips: z.ZodArray<z.ZodString>;
     cnameTarget: z.ZodOptional<z.ZodString>;
+    lbMode: z.ZodOptional<z.ZodEnum<{
+        round_robin: "round_robin";
+        failover: "failover";
+        weighted: "weighted";
+    }>>;
     deleted: z.ZodOptional<z.ZodBoolean>;
 }, z.core.$strip>;
 declare const cfdmSyncBindingsBodySchema: z.ZodObject<{
@@ -2231,6 +2265,11 @@ declare const cfdmSyncBindingsBodySchema: z.ZodObject<{
         hostname: z.ZodString;
         ips: z.ZodArray<z.ZodString>;
         cnameTarget: z.ZodOptional<z.ZodString>;
+        lbMode: z.ZodOptional<z.ZodEnum<{
+            round_robin: "round_robin";
+            failover: "failover";
+            weighted: "weighted";
+        }>>;
         deleted: z.ZodOptional<z.ZodBoolean>;
     }, z.core.$strip>>;
     fullSync: z.ZodOptional<z.ZodBoolean>;
@@ -2412,4 +2451,4 @@ declare const ingestAuditEventSchema: z.ZodObject<{
 }, z.core.$strip>;
 type IngestAuditEvent = z.infer<typeof ingestAuditEventSchema>;
 
-export { AUDIT_SEVERITIES, AUDIT_SOURCE_APPS, AUDIT_TARGET_TYPES, type AppSettingsPatch, type AppSwitcherConfig, type AppSwitcherEntry, type AuditListQuery, type AuditLogEntry, type AuditSeverity, type AuditSourceApp, type AuditTargetType, type BulkUpdateDomainsInput, CERT_ERROR, CERT_EXPIRED, CERT_MONITORING_VALUES, CERT_MONITOR_AUTO, CERT_MONITOR_REQUIRED, CERT_MONITOR_SKIPPED, CERT_OK, CERT_UNKNOWN, CERT_WARNING, type CertMonitoring, type Certificate, type CfDnsRecord, type CfHealthCheck, type CfZone, type CfdmBindingSyncItem, type ChangeDomainInput, type ChangeIpInput, type CreateDnsRecordInput, type CreateDnsRecordPayload, type CreateDomainInput, type CreateDomainMonitorInput, type CreateGroupInput, type CreateOriginHealthCheckInput, type CreateServiceBindingInput, type CreateServiceGroupInput, type CreateServiceInput, type CreateServiceNodeInput, type CreateServiceWithConfigInput, type CreateSubdomainInput, type DnsRecord, type Domain, type DomainEnvironment, type DomainListItem, type DomainMonitor, type DomainMonitorResult, type DomainMonitorType, type Group, type GroupWithStats, HEALTH_CHECK_AGGREGATES, HEALTH_CHECK_PROVIDERS, HEALTH_KV_CURSOR_KEY, HEALTH_KV_RESULTS_KEY, HEALTH_KV_TARGETS_KEY, HEALTH_PROBE_BATCH, HEALTH_PROBE_CONCURRENCY, HEALTH_PROBE_KV_TITLE, HEALTH_PROBE_SCRIPT_NAME, HEALTH_STATUS_PROVIDERS, type HealthCheckAggregate, type HealthCheckConfig, type HealthCheckProvider, type HealthCheckScope, type HealthCheckTarget, type HealthCheckType, type HealthProbeLog, type HealthProbeResultItem, type HealthProbeResultsDoc, type HealthProbeTargetItem, type HealthProbeTargetsDoc, type HealthStatusProvider, type HealthStatusQuery, type HealthWorkerStatus, type IngestAuditEvent, type IpHealthState, type IpHealthStatus, type JwtClaims, type LbMode, type LoginInput, type LoginRequest, type LoginResponse, type NodeHealthState, type NotificationLog, type OriginHealthCheck, type OriginHealthCheckRecord, type ParsedFqdn, type PatchDnsRecordPayload, type ReorderServicesInput, SYNC_CONFLICT, SYNC_ERROR, SYNC_PENDING_DELETE, SYNC_PENDING_PUSH, SYNC_SYNCED, type Service, type ServiceBinding, type ServiceBindingView, type ServiceCertificateRow, type ServiceDomainBinding, type ServiceDomainBindingView, type ServiceGroup, type ServiceGroupView, type ServiceGroupsResponse, type ServiceIpHealth, type ServiceNode, type ServiceNodeRecord, type ServiceOverview, type ServiceView, type Subdomain, type SubdomainRecord, type SyncJob, type ToggleEnabledInput, type ToggleServiceIpInput, type UpdateDomainInput, type UpdateServiceConfigInput, type UpdateServiceGroupInput, type UpdateServiceNodeInput, type UpdateSubdomainInput, ValidationError, type VpsTrackerEvent, aggregateHealthOk, appSettingsPatchSchema, appSwitcherConfigSchema, appSwitcherEntrySchema, appSwitcherIconSchema, auditListQuerySchema, auditLogEntrySchema, auditSeveritySchema, auditSourceAppSchema, auditTargetTypeSchema, bindingToFqdn, bulkUpdateDomainsSchema, certMonitoringSchema, certStatusFromExpiry, certificateSchema, cfdmBindingSyncItemSchema, cfdmSyncBindingsBodySchema, changeDomainSchema, changeIpSchema, clampGlobalpingLimit, createDnsRecordSchema, createDomainMonitorSchema, createDomainSchema, createGroupSchema, createOriginHealthCheckSchema, createServiceBindingSchema, createServiceGroupSchema, createServiceNodeSchema, createServiceSchema, createServiceWithConfigSchema, createSubdomainSchema, derivePrimaryProvider, dnsNameToSubdomainLabel, dnsRecordNamesMatch, dnsRecordSchema, domainEnvironmentSchema, domainListItemSchema, domainMonitorResultSchema, domainMonitorSchema, domainMonitorTypeSchema, domainSchema, fqdnToDisplay, groupSchema, groupWithStatsSchema, healthCheckAggregateSchema, healthCheckConfigSchema, healthCheckProviderSchema, healthCheckProvidersSchema, healthCheckScopeSchema, healthCheckTypeSchema, healthProbeLogSchema, healthStatusProviderSchema, healthStatusQuerySchema, ingestAuditEventSchema, ipHealthStateSchema, ipHealthStatusSchema, isIpLiteral, isValidIpv4, lbModeSchema, loginSchema, nodeHealthStateSchema, normalizeDnsRecordName, normalizeProbeProvider, normalizeStatusProvider, notificationLogSchema, originHealthCheckSchema, parseFqdn, parseGlobalpingLocations, parseHealthAggregate, parseHealthProviders, reorderServicesSchema, serializeHealthProviders, serviceBindingSchema, serviceCertificateRowSchema, serviceDomainBindingSchema, serviceGroupSchema, serviceGroupTypeSchema, serviceGroupViewSchema, serviceGroupsResponseSchema, serviceIpHealthSchema, serviceNodeSchema, serviceSchema, serviceViewSchema, shouldMonitorService, subdomainLabelToFqdn, subdomainSchema, targetHasProvider, targetProviders, toggleEnabledSchema, toggleServiceIpSchema, uniqueHealthProviders, updateDomainGroupSchema, updateDomainSchema, updateServiceConfigSchema, updateServiceGroupSchema, updateServiceNodeSchema, updateSubdomainSchema, validateDnsRecord, vpsTrackerEventSchema };
+export { AUDIT_SEVERITIES, AUDIT_SOURCE_APPS, AUDIT_TARGET_TYPES, type AppSettingsPatch, type AppSwitcherConfig, type AppSwitcherEntry, type AuditListQuery, type AuditLogEntry, type AuditSeverity, type AuditSourceApp, type AuditTargetType, type BulkUpdateDomainsInput, CERT_ERROR, CERT_EXPIRED, CERT_MONITORING_VALUES, CERT_MONITOR_AUTO, CERT_MONITOR_REQUIRED, CERT_MONITOR_SKIPPED, CERT_OK, CERT_UNKNOWN, CERT_WARNING, type CertMonitoring, type Certificate, type CfDnsRecord, type CfHealthCheck, type CfZone, type CfdmBindingSyncItem, type ChangeDomainInput, type ChangeIpInput, type CreateDnsRecordInput, type CreateDnsRecordPayload, type CreateDomainInput, type CreateDomainMonitorInput, type CreateGroupInput, type CreateOriginHealthCheckInput, type CreateServiceBindingInput, type CreateServiceGroupInput, type CreateServiceInput, type CreateServiceNodeInput, type CreateServiceWithConfigInput, type CreateSubdomainInput, type DnsRecord, type Domain, type DomainEnvironment, type DomainListItem, type DomainMonitor, type DomainMonitorResult, type DomainMonitorType, type FailoverLogEntry, type Group, type GroupWithStats, HEALTH_CHECK_AGGREGATES, HEALTH_CHECK_PROVIDERS, HEALTH_KV_CURSOR_KEY, HEALTH_KV_RESULTS_KEY, HEALTH_KV_TARGETS_KEY, HEALTH_PROBE_BATCH, HEALTH_PROBE_CONCURRENCY, HEALTH_PROBE_KV_TITLE, HEALTH_PROBE_SCRIPT_NAME, HEALTH_STATUS_PROVIDERS, type HealthCheckAggregate, type HealthCheckConfig, type HealthCheckProvider, type HealthCheckScope, type HealthCheckTarget, type HealthCheckType, type HealthProbeLog, type HealthProbeResultItem, type HealthProbeResultsDoc, type HealthProbeTargetItem, type HealthProbeTargetsDoc, type HealthStatusProvider, type HealthStatusQuery, type HealthWorkerStatus, type IngestAuditEvent, type IpHealthState, type IpHealthStatus, type JwtClaims, type LbMode, type LoginInput, type LoginRequest, type LoginResponse, type NodeHealthState, type NotificationLog, type OriginHealthCheck, type OriginHealthCheckRecord, type ParsedFqdn, type PatchDnsRecordPayload, type ReorderServicesInput, SYNC_CONFLICT, SYNC_ERROR, SYNC_PENDING_DELETE, SYNC_PENDING_PUSH, SYNC_SYNCED, type Service, type ServiceBinding, type ServiceBindingView, type ServiceCertificateRow, type ServiceDomainBinding, type ServiceDomainBindingView, type ServiceGroup, type ServiceGroupView, type ServiceGroupsResponse, type ServiceIpHealth, type ServiceNode, type ServiceNodeRecord, type ServiceOverview, type ServiceView, type Subdomain, type SubdomainRecord, type SyncJob, type ToggleEnabledInput, type ToggleServiceIpInput, type UpdateDomainInput, type UpdateServiceConfigInput, type UpdateServiceGroupInput, type UpdateServiceNodeInput, type UpdateSubdomainInput, ValidationError, type VpsTrackerEvent, aggregateHealthOk, appSettingsPatchSchema, appSwitcherConfigSchema, appSwitcherEntrySchema, appSwitcherIconSchema, auditListQuerySchema, auditLogEntrySchema, auditSeveritySchema, auditSourceAppSchema, auditTargetTypeSchema, bindingToFqdn, bulkUpdateDomainsSchema, certMonitoringSchema, certStatusFromExpiry, certificateSchema, cfdmBindingSyncItemSchema, cfdmSyncBindingsBodySchema, changeDomainSchema, changeIpSchema, clampGlobalpingLimit, createDnsRecordSchema, createDomainMonitorSchema, createDomainSchema, createGroupSchema, createOriginHealthCheckSchema, createServiceBindingSchema, createServiceGroupSchema, createServiceNodeSchema, createServiceSchema, createServiceWithConfigSchema, createSubdomainSchema, derivePrimaryProvider, dnsNameToSubdomainLabel, dnsRecordNamesMatch, dnsRecordSchema, domainEnvironmentSchema, domainListItemSchema, domainMonitorResultSchema, domainMonitorSchema, domainMonitorTypeSchema, domainSchema, failoverLogSchema, fqdnToDisplay, groupSchema, groupWithStatsSchema, healthCheckAggregateSchema, healthCheckConfigSchema, healthCheckProviderSchema, healthCheckProvidersSchema, healthCheckScopeSchema, healthCheckTypeSchema, healthProbeLogSchema, healthStatusProviderSchema, healthStatusQuerySchema, ingestAuditEventSchema, ipHealthStateSchema, ipHealthStatusSchema, isIpLiteral, isValidIpv4, lbModeSchema, loginSchema, nodeHealthStateSchema, normalizeDnsRecordName, normalizeProbeProvider, normalizeStatusProvider, notificationLogSchema, originHealthCheckSchema, parseFqdn, parseGlobalpingLocations, parseHealthAggregate, parseHealthProviders, reorderServicesSchema, serializeHealthProviders, serviceBindingSchema, serviceCertificateRowSchema, serviceDomainBindingSchema, serviceGroupSchema, serviceGroupTypeSchema, serviceGroupViewSchema, serviceGroupsResponseSchema, serviceIpHealthSchema, serviceNodeSchema, serviceSchema, serviceViewSchema, shouldMonitorService, subdomainLabelToFqdn, subdomainSchema, targetHasProvider, targetProviders, toggleEnabledSchema, toggleServiceIpSchema, uniqueHealthProviders, updateDomainGroupSchema, updateDomainSchema, updateServiceConfigSchema, updateServiceGroupSchema, updateServiceNodeSchema, updateSubdomainSchema, validateDnsRecord, vpsTrackerEventSchema };

packages/shared/dist/index.js

@@ -19,7 +19,10 @@ var CERT_MONITORING_VALUES = [
 ];
 
 // src/validators.ts
-var NAME_RE = /^(@|\*|[a-zA-Z0-9_]([a-zA-Z0-9_-]*[a-zA-Z0-9_])?(\.[a-zA-Z0-9_]([a-zA-Z0-9_-]*[a-zA-Z0-9_])?)*)$/;
+var LABEL_RE = "[a-zA-Z0-9_](?:[a-zA-Z0-9_-]*[a-zA-Z0-9_])?";
+var NAME_RE = new RegExp(
+  `^(@|\\*|(\\*\\.)?${LABEL_RE}(?:\\.${LABEL_RE})*)$`
+);
 var IPV4_RE = /^((25[0-5]|2[0-4]\d|[01]?\d\d?)\.){3}(25[0-5]|2[0-4]\d|[01]?\d\d?)$/;
 var IPV6_RE = /^([0-9a-fA-F]{0,4}:){2,7}[0-9a-fA-F]{0,4}$/;
 var ALLOWED_TYPES = ["A", "AAAA", "CNAME", "TXT", "MX", "NS", "SRV", "CAA"];
@@ -106,6 +109,7 @@ function dnsNameToSubdomainLabel(recordName, zoneName) {
     const prefix = rn.slice(0, rn.length - zoneSuffix.length);
     return prefix || "@";
   }
+  if (rn.startsWith("*.")) return rn;
   if (!rn.includes(".")) return rn;
   return null;
 }
@@ -322,6 +326,15 @@ var healthProbeLogSchema = z.object({
   error: z.string().nullable(),
   checked_at: z.string()
 });
+var failoverLogSchema = z.object({
+  id: z.number(),
+  service_id: z.number(),
+  binding_id: z.number(),
+  fqdn: z.string(),
+  ip: z.string(),
+  action: z.enum(["added", "removed"]),
+  created_at: z.string()
+});
 var groupSchema = z.object({
   id: z.number(),
   name: z.string(),
@@ -399,7 +412,8 @@ var serviceDomainBindingSchema = z.object({
   health_check_providers: healthCheckProvidersSchema.catch(["local"]),
   health_check_aggregate: healthCheckAggregateSchema.catch("majority"),
   cert_monitoring: certMonitoringSchema.default("auto"),
-  sync_status: z.string().nullable().default(null)
+  sync_status: z.string().nullable().default(null),
+  active_ips: z.array(z.string()).default([])
 }).transform((binding) => ({
   ...binding,
   target_ips: binding.target_ips && binding.target_ips.length > 0 ? binding.target_ips : binding.target_ip ? [binding.target_ip] : [],
@@ -836,6 +850,8 @@ var cfdmBindingSyncItemSchema = z3.object({
   ips: z3.array(z3.string()),
   /** CNAME-цель (FQDN), если binding — CNAME; для матчинга в VPS Tracker. */
   cnameTarget: z3.string().optional(),
+  /** HA-режим binding (fallback — service group). Optional для старых payload. */
+  lbMode: z3.enum(["round_robin", "failover", "weighted"]).optional(),
   deleted: z3.boolean().optional()
 });
 var cfdmSyncBindingsBodySchema = z3.object({
@@ -1019,6 +1035,7 @@ export {
   domainMonitorSchema,
   domainMonitorTypeSchema,
   domainSchema,
+  failoverLogSchema,
   fqdnToDisplay,
   groupSchema,
   groupWithStatsSchema,

packages/shared/src/integration-vps-tracker.ts

@@ -11,6 +11,8 @@ export const cfdmBindingSyncItemSchema = z.object({
   ips: z.array(z.string()),
   /** CNAME-цель (FQDN), если binding — CNAME; для матчинга в VPS Tracker. */
   cnameTarget: z.string().optional(),
+  /** HA-режим binding (fallback — service group). Optional для старых payload. */
+  lbMode: z.enum(["round_robin", "failover", "weighted"]).optional(),
   deleted: z.boolean().optional(),
 });
 

packages/shared/src/schemas.ts

@@ -100,6 +100,18 @@ export const healthProbeLogSchema = z.object({
 
 export type HealthProbeLog = z.infer<typeof healthProbeLogSchema>
 
+export const failoverLogSchema = z.object({
+  id: z.number(),
+  service_id: z.number(),
+  binding_id: z.number(),
+  fqdn: z.string(),
+  ip: z.string(),
+  action: z.enum(['added', 'removed']),
+  created_at: z.string(),
+})
+
+export type FailoverLogEntry = z.infer<typeof failoverLogSchema>
+
 export const groupSchema = z.object({
   id: z.number(),
   name: z.string(),
@@ -184,6 +196,7 @@ export const serviceDomainBindingSchema = z
     health_check_aggregate: healthCheckAggregateSchema.catch('majority'),
     cert_monitoring: certMonitoringSchema.default('auto'),
     sync_status: z.string().nullable().default(null),
+    active_ips: z.array(z.string()).default([]),
   })
   .transform((binding) => ({
     ...binding,

packages/shared/src/subdomain.ts

@@ -19,6 +19,7 @@ export function dnsNameToSubdomainLabel(
     return prefix || "@";
   }
 
+  if (rn.startsWith("*.")) return rn;
   if (!rn.includes(".")) return rn;
 
   return null;

packages/shared/src/types.ts

@@ -190,6 +190,7 @@ export interface ServiceDomainBindingView {
   fqdn: string;
   record_type: "A" | "CNAME";
   target_ips: string[];
+  active_ips: string[];
   target_ip_weights: Record<string, number>;
   target_ip_priorities: Record<string, number>;
   target_cname: string | null;

packages/shared/src/validators.ts

@@ -5,8 +5,11 @@ import {
 } from "./constants.js";
 import type { ServiceGroup } from "./types.js";
 
-const NAME_RE =
-  /^(@|\*|[a-zA-Z0-9_]([a-zA-Z0-9_-]*[a-zA-Z0-9_])?(\.[a-zA-Z0-9_]([a-zA-Z0-9_-]*[a-zA-Z0-9_])?)*)$/;
+const LABEL_RE = "[a-zA-Z0-9_](?:[a-zA-Z0-9_-]*[a-zA-Z0-9_])?";
+/** Apex `@`, zone `*`, labels, or nested wildcard (`*.ndns`, `*.ndns.shnt.top`). */
+const NAME_RE = new RegExp(
+  `^(@|\\*|(\\*\\.)?${LABEL_RE}(?:\\.${LABEL_RE})*)$`,
+);
 const IPV4_RE =
   /^((25[0-5]|2[0-4]\d|[01]?\d\d?)\.){3}(25[0-5]|2[0-4]\d|[01]?\d\d?)$/;
 const IPV6_RE = /^([0-9a-fA-F]{0,4}:){2,7}[0-9a-fA-F]{0,4}$/;

packages/shared/test/subdomain.test.ts

@@ -23,6 +23,13 @@ describe("normalizeDnsRecordName", () => {
     expect(normalizeDnsRecordName("@", ZONE)).toBe("rkns.top");
     expect(normalizeDnsRecordName("rkns.top", ZONE)).toBe("rkns.top");
   });
+
+  it("normalizes nested wildcard relative name to FQDN", () => {
+    expect(normalizeDnsRecordName("*.mdns", ZONE)).toBe("*.mdns.rkns.top");
+    expect(normalizeDnsRecordName("*.mdns.rkns.top", ZONE)).toBe(
+      "*.mdns.rkns.top",
+    );
+  });
 });
 
 describe("dnsRecordNamesMatch", () => {
@@ -34,10 +41,19 @@ describe("dnsRecordNamesMatch", () => {
   it("does not match different hosts", () => {
     expect(dnsRecordNamesMatch("de", "mhome.rkns.top", ZONE)).toBe(false);
   });
+
+  it("matches nested wildcard relative name and FQDN", () => {
+    expect(dnsRecordNamesMatch("*.mdns", "*.mdns.rkns.top", ZONE)).toBe(true);
+  });
 });
 
 describe("dnsNameToSubdomainLabel", () => {
   it("extracts label from FQDN", () => {
     expect(dnsNameToSubdomainLabel("de.rkns.top", ZONE)).toBe("de");
   });
+
+  it("keeps nested wildcard relative names", () => {
+    expect(dnsNameToSubdomainLabel("*.mdns", ZONE)).toBe("*.mdns");
+    expect(dnsNameToSubdomainLabel("*.mdns.rkns.top", ZONE)).toBe("*.mdns");
+  });
 });

packages/shared/test/validators.test.ts

@@ -0,0 +1,41 @@
+import { describe, expect, it } from "vitest";
+import { validateDnsRecord, ValidationError } from "../src/validators.js";
+
+function assertValidName(name: string): void {
+  expect(() =>
+    validateDnsRecord("A", name, "1.2.3.4", 1, false),
+  ).not.toThrow();
+}
+
+function assertInvalidName(name: string): void {
+  expect(() => validateDnsRecord("A", name, "1.2.3.4", 1, false)).toThrow(
+    ValidationError,
+  );
+  expect(() => validateDnsRecord("A", name, "1.2.3.4", 1, false)).toThrow(
+    `invalid record name: ${name}`,
+  );
+}
+
+describe("validateDnsRecord name", () => {
+  it("accepts apex and zone wildcard", () => {
+    assertValidName("@");
+    assertValidName("*");
+  });
+
+  it("accepts regular labels and FQDN", () => {
+    assertValidName("ndns");
+    assertValidName("ndns.shnt.top");
+  });
+
+  it("accepts nested wildcard relative name and FQDN", () => {
+    assertValidName("*.ndns");
+    assertValidName("*.ndns.shnt.top");
+    assertValidName("*.mdns");
+    assertValidName("*.mdns.rkns.top");
+  });
+
+  it("rejects wildcard not as leftmost label", () => {
+    assertInvalidName("foo.*.bar");
+    assertInvalidName("ndns.*");
+  });
+});