Merge pull request #116 from Routstr/mint-recovery-probe-restore

wallet: probe mint reachability, per-mint recovery gating

src/daemon/http/index.ts

@@ -572,6 +572,19 @@ export function createDaemonRequestHandler(deps: {
       return;
     }
 
+    if (req.method === "POST" && url.pathname === "/wallet/recover/operations") {
+      await respond(res, async () => {
+        if (!deps.walletClient.recoverStuckOperations) {
+          throw new CocodHttpError(
+            501,
+            "Stuck operation recovery is not supported by this wallet client.",
+          );
+        }
+        return { output: await deps.walletClient.recoverStuckOperations() };
+      });
+      return;
+    }
+
     if (req.method === "POST" && url.pathname === "/wallet/receive/cashu") {
       await respond(res, async () => {
         const body = await readJsonBody(req);

src/daemon/http/wallet-recovery.test.ts

@@ -47,3 +47,53 @@ describe("POST /wallet/recover validation", () => {
     expect(recoverMintQuotes).toHaveBeenCalledTimes(1);
   });
 });
+
+async function recoverOperations() {
+  const recoverStuckOperations = mock(async () => ({
+    attempted: 1,
+    busy: 0,
+    skipped: 0,
+    failed: 0,
+    skippedMints: {},
+  }));
+  const handler = createDaemonRequestHandler({ walletClient: { recoverStuckOperations } } as never);
+  const req = new EventEmitter() as any;
+  Object.assign(req, { method: "POST", url: "/wallet/recover/operations", headers: { host: "localhost" } });
+  const res = {
+    status: 0, body: "",
+    writeHead(status: number) { this.status = status; },
+    end(chunk: string) { this.body = chunk; },
+  };
+  setImmediate(() => {
+    req.emit("data", Buffer.from("{}"));
+    req.emit("end");
+  });
+  await handler(req, res as never);
+  return { res, recoverStuckOperations };
+}
+
+describe("POST /wallet/recover/operations", () => {
+  it("drives stuck-operation recovery and returns the summary", async () => {
+    const { res, recoverStuckOperations } = await recoverOperations();
+    expect(res.status).toBe(200);
+    expect(recoverStuckOperations).toHaveBeenCalledTimes(1);
+    expect(JSON.parse(res.body).output).toMatchObject({ attempted: 1, busy: 0 });
+  });
+
+  it("returns 501 when the wallet client does not support it", async () => {
+    const handler = createDaemonRequestHandler({ walletClient: {} } as never);
+    const req = new EventEmitter() as any;
+    Object.assign(req, { method: "POST", url: "/wallet/recover/operations", headers: { host: "localhost" } });
+    const res = {
+      status: 0, body: "",
+      writeHead(status: number) { this.status = status; },
+      end(chunk: string) { this.body = chunk; },
+    };
+    setImmediate(() => {
+      req.emit("data", Buffer.from("{}"));
+      req.emit("end");
+    });
+    await handler(req, res as never);
+    expect(res.status).toBe(501);
+  });
+});

src/daemon/wallet/coco-client.test.ts

@@ -14,6 +14,7 @@ import {
   assertLegacyCocodNotRunning,
   claimLegacyCocodPidFile,
   createCocoClient,
+  createRecoveryGate,
   createRunQueue,
   DEFAULT_TRUSTED_MINT_URLS,
   failExpiredMintQuoteIfUnpaid,
@@ -680,6 +681,29 @@ describe("settleExpiredMintQuotes", () => {
     expect(observePendingOperation).not.toHaveBeenCalled();
     expect(failPendingOperation).not.toHaveBeenCalled();
   });
+
+  it("skips quotes at probe-unreachable mints without spending the budget", async () => {
+    const ops = [
+      pendingMintOp({ id: "dead-1", mintUrl: "https://dead.example.com" }),
+      pendingMintOp({ id: "dead-2", mintUrl: "https://dead.example.com/" }),
+      pendingMintOp({ id: "live-1", mintUrl: "https://live.example.com" }),
+    ];
+    const { source, observePendingOperation, failPendingOperation } =
+      fakeSource(ops);
+
+    const result = await settleExpiredMintQuotes(
+      source,
+      NOW_MS,
+      undefined,
+      { unreachableMints: new Set(["https://dead.example.com"]) },
+    );
+
+    expect(result).toEqual({ failed: 1, leftForRecovery: 0, unobserved: 2 });
+    // Only the reachable mint was asked anything.
+    expect(observePendingOperation).toHaveBeenCalledTimes(1);
+    expect(observePendingOperation.mock.calls[0]?.[0]).toBe("live-1");
+    expect(failPendingOperation).toHaveBeenCalledTimes(1);
+  });
 });
 
 describe("settlePendingMintQuotes", () => {
@@ -1536,7 +1560,7 @@ describe("runMintQuoteRecovery", () => {
 
   it("skips operations whose earlier recovery is still in flight", async () => {
     const outstanding = new Map<string, Promise<unknown>>([
-      ["op-1", new Promise(() => {})],
+      ["mint:op-1", new Promise(() => {})],
     ]);
     const { source, finalize, observePendingOperation } = fakeSource(
       [mintOp()],
@@ -1567,14 +1591,14 @@ describe("runMintQuoteRecovery", () => {
     });
 
     expect(first).toMatchObject({ retryable: 1, recovered: 0 });
-    expect(outstanding.has("op-1")).toBe(true);
+    expect(outstanding.has("mint:op-1")).toBe(true);
     // The abandoned mint request must not be retried underneath.
     expect(second).toMatchObject({ busy: 1, checked: 0 });
   });
 
   it("does not re-open a failed operation whose recovery is in flight", async () => {
     const outstanding = new Map<string, Promise<unknown>>([
-      ["op-1", new Promise(() => {})],
+      ["mint:op-1", new Promise(() => {})],
     ]);
     const { source, reopenFailedOperation } = fakeSource(
       [mintOp({ state: "failed" })],
@@ -1629,8 +1653,78 @@ describe("runMintQuoteRecovery", () => {
     });
 
     expect(first).toMatchObject({ retryable: 1, checked: 1 });
-    expect(outstanding.has("op-1")).toBe(true);
+    expect(outstanding.has("mint:op-1")).toBe(true);
     expect(second).toMatchObject({ busy: 1, checked: 0 });
     expect(finalize).not.toHaveBeenCalled();
   });
 });
+
+describe("createRecoveryGate", () => {
+  const HEALTHY = "https://healthy.example.com";
+  const STUCK = "https://stuck.example.com";
+
+  function settled(promise: Promise<unknown>): Promise<boolean> {
+    return Promise.race([
+      promise.then(() => true, () => true),
+      new Promise<boolean>((resolve) => setTimeout(() => resolve(false), 25)),
+    ]);
+  }
+
+  it("lets a mint without stuck operations proceed while another mint recovers", async () => {
+    const gate = createRecoveryGate();
+    gate.publishStuckMints(new Set([STUCK]));
+    // Recovery is still running (complete() never called), yet the healthy
+    // mint must not be blocked by the stuck one.
+    await gate.waitForRecovery(HEALTHY);
+  });
+
+  it("holds a mint with stuck operations until recovery completes", async () => {
+    const gate = createRecoveryGate();
+    gate.publishStuckMints(new Set([STUCK]));
+    const waiting = gate.waitForRecovery(STUCK);
+    expect(await settled(waiting)).toBe(false);
+    gate.complete();
+    await waiting;
+  });
+
+  it("waits for the stuck-mint enumeration before deciding", async () => {
+    const gate = createRecoveryGate();
+    const waiting = gate.waitForRecovery(HEALTHY);
+    expect(await settled(waiting)).toBe(false);
+    gate.publishStuckMints(new Set([STUCK]));
+    await waiting;
+  });
+
+  it("holds callers without a target mint until recovery completes", async () => {
+    const gate = createRecoveryGate();
+    gate.publishStuckMints(new Set([STUCK]));
+    const waiting = gate.waitForRecovery();
+    expect(await settled(waiting)).toBe(false);
+    gate.complete();
+    await waiting;
+  });
+
+  it("poisons every caller after a recovery failure", async () => {
+    const gate = createRecoveryGate();
+    gate.publishStuckMints(new Set([STUCK]));
+    gate.fail("disk exploded");
+    await expect(gate.waitForRecovery(HEALTHY)).rejects.toThrow(
+      "Wallet is not ready: disk exploded",
+    );
+    await expect(gate.waitForRecovery(STUCK)).rejects.toThrow(
+      "Wallet is not ready: disk exploded",
+    );
+    await expect(gate.waitForRecovery()).rejects.toThrow(
+      "Wallet is not ready: disk exploded",
+    );
+  });
+
+  it("falls back to the global gate for unparseable mint URLs", async () => {
+    const gate = createRecoveryGate();
+    gate.publishStuckMints(new Set([STUCK]));
+    const waiting = gate.waitForRecovery("not a url");
+    expect(await settled(waiting)).toBe(false);
+    gate.complete();
+    await waiting;
+  });
+});

src/daemon/wallet/coco-client.ts

@@ -1,3 +1,4 @@
+import { recoveryKey, trackRecovery, drainRecoveryWork, waitForRecoveryWork, createRecoveryDisposer, type RecoveryWork } from "./recovery-work";
 import {
   Manager,
   OperationInProgressError,
@@ -42,6 +43,13 @@ import {
   selectMintQuotesForRecovery,
   type MintQuoteRecoveryCandidate,
 } from "./mint-quote-recovery";
+import {
+  collectStuckOperations,
+  probeMintReachability,
+  runTargetedRecovery,
+  type SendRecoveryService,
+  type StuckOperation,
+} from "./recovery-probe";
 import {
   clearInterruptedReceiveReservations,
   deleteReceiveTokenReservation,
@@ -724,6 +732,7 @@ const EXPIRED_MINT_OBSERVATION_DEADLINE_MS = 15_000;
 
 /** Rejects when `timeoutMs` elapses before `promise` settles. */
 function withTimeout<T>(promise: Promise<T>, timeoutMs: number): Promise<T> {
+  if (timeoutMs === Infinity) return promise;
   let timer: ReturnType<typeof setTimeout> | undefined;
   const timeout = new Promise<never>((_resolve, reject) => {
     timer = setTimeout(
@@ -838,6 +847,7 @@ export async function settleExpiredMintQuotes(
   source: ExpiredMintQuoteSource,
   nowMs: number,
   deadlineMs: number = EXPIRED_MINT_OBSERVATION_DEADLINE_MS,
+  options: { unreachableMints?: Set<string>; outstanding?: RecoveryWork; shouldStop?: () => boolean } = {},
 ): Promise<ExpiredMintSettlement> {
   const pendingMints = await source.ops.mint.listPending();
   const selection = selectCleanupOperations({
@@ -858,6 +868,25 @@ export async function settleExpiredMintQuotes(
 
   const startedAt = Date.now();
   for (const op of candidates) {
+    if (options.shouldStop?.() || options.outstanding?.has(recoveryKey("mint", op.id))) {
+      settlement.unobserved++; continue;
+    }
+    // A mint the startup probe already found unreachable cannot answer an
+    // observation either; skip it without spending the shared wall-clock
+    // budget, leaving the quote pending for a later startup.
+    if (options.unreachableMints) {
+      let mintUrl = op.mintUrl;
+      try {
+        mintUrl = normalizeMintUrl(op.mintUrl);
+      } catch {
+        // Malformed persisted URL: probe keys are raw for those, and the
+        // observation below would fail anyway, landing in `unobserved`.
+      }
+      if (options.unreachableMints.has(mintUrl)) {
+        settlement.unobserved++;
+        continue;
+      }
+    }
     const remainingMs = deadlineMs - (Date.now() - startedAt);
     if (remainingMs <= 0) {
       const skipped =
@@ -872,11 +901,12 @@ export async function settleExpiredMintQuotes(
       break;
     }
 
-    const check = await failExpiredMintQuoteIfUnpaid(
-      source.mintOperationService,
-      op.id,
-      remainingMs,
-    );
+    // Track the complete observation/failure chain, not just its bounded wait.
+    const work = failExpiredMintQuoteIfUnpaid(source.mintOperationService, op.id, Infinity);
+    if (options.outstanding) trackRecovery(options.outstanding, recoveryKey("mint", op.id), work);
+    const check = await waitForRecoveryWork(work, remainingMs).catch(error => ({
+      outcome: "unobserved" as const, error, category: undefined,
+    }));
     if (check.outcome === "failed") {
       settlement.failed++;
     } else if (check.outcome === "leftForRecovery") {
@@ -937,6 +967,7 @@ export interface MintQuoteRecoverySource {
 }
 
 export interface MintQuoteRecoveryOptions {
+  shouldStop?: () => boolean;
   /** Target only these operation ids (may include failed operations). */
   operationIds?: string[];
   /** Per-quote budget for observing the mint and finalizing the operation. */
@@ -948,7 +979,7 @@ export interface MintQuoteRecoveryOptions {
    */
   includeFailed?: boolean;
   /**
-   * In-flight recovery work keyed by operation id, shared across runs.
+   * In-flight recovery work keyed by family:id (mint:<operation id>), shared across runs.
    * withTimeout does not cancel the underlying request, so a timed-out quote
    * check or finalize must keep blocking a retry until it actually settles.
    */
@@ -987,9 +1018,9 @@ export interface MintQuoteRecoveryResult {
  * underneath work that outlived its timeout. A rejected task never breaks the
  * chain for the next one.
  */
-export function createRunQueue(): <T>(run: () => Promise<T>) => Promise<T> {
+export function createRunQueue(): (<T>(run: () => Promise<T>) => Promise<T>) & { drain(): Promise<void> } {
   let tail: Promise<unknown> = Promise.resolve();
-  return <T>(run: () => Promise<T>): Promise<T> => {
+  const enqueue = <T>(run: () => Promise<T>): Promise<T> => {
     const result = tail.then(run, run);
     tail = result.then(
       () => undefined,
@@ -997,6 +1028,7 @@ export function createRunQueue(): <T>(run: () => Promise<T>) => Promise<T> {
     );
     return result;
   };
+  return Object.assign(enqueue, { drain: async () => { await tail; } });
 }
 
 /** Per-quote budget for the mint round-trip during explicit recovery. */
@@ -1064,21 +1096,8 @@ export async function runMintQuoteRecovery(
   /** coco's fail-fast operation lock rejected the call: another holder exists. */
   const isInProgress = (error: unknown) =>
     error instanceof Error && error.name === "OperationInProgressError";
-  /**
-   * Register in-flight work for an operation. Entries are cleared only once the
-   * work actually settles (withTimeout does not cancel the request behind it),
-   * so a timed-out call keeps blocking a retry. The identity check stops a late
-   * settlement from clearing a newer entry for the same operation.
-   */
-  const track = (operationId: string, work: Promise<unknown>) => {
-    outstanding.set(operationId, work);
-    const clear = () => {
-      if (outstanding.get(operationId) === work) {
-        outstanding.delete(operationId);
-      }
-    };
-    void work.then(clear, clear);
-  };
+  const track = (operationId: string, work: Promise<unknown>) =>
+    trackRecovery(outstanding, recoveryKey("mint", operationId), work);
 
   let targets: MintQuoteRecoveryCandidate[];
   if (options.operationIds && options.operationIds.length > 0) {
@@ -1108,8 +1127,9 @@ export async function runMintQuoteRecovery(
   });
 
   for (const op of failed) {
+    if (options.shouldStop?.()) break;
     const label = `Mint quote ${op.quoteId ?? op.id} at ${op.mintUrl}`;
-    if (outstanding.has(op.id)) {
+    if (outstanding.has(recoveryKey("mint", op.id))) {
       result.busy++;
       onProgress?.(`${label}: an earlier recovery is still running; skipped`);
       continue;
@@ -1137,13 +1157,16 @@ export async function runMintQuoteRecovery(
     await recoverOne(op);
   }
 
-  for (const op of pending) await recoverOne(op);
+  for (const op of pending) {
+    if (options.shouldStop?.()) break;
+    await recoverOne(op);
+  }
 
   return result;
 
   async function recoverOne(op: MintQuoteRecoveryCandidate): Promise<void> {
     const label = `Mint quote ${op.quoteId ?? op.id} at ${op.mintUrl}`;
-    if (outstanding.has(op.id)) {
+    if (outstanding.has(recoveryKey("mint", op.id))) {
       result.busy++;
       onProgress?.(`${label}: an earlier recovery is still running; skipped`);
       return;
@@ -1298,6 +1321,7 @@ export interface PendingMintSweepOptions {
   deadlineMs?: number;
   checkTimeoutMs?: number;
   state?: PendingMintSweepState;
+  shouldStop?: () => boolean;
 }
 
 type PendingMintOutcome = "unreachable" | "other";
@@ -1319,21 +1343,21 @@ export async function settlePendingMintQuotes(
   const ordered = [...pending.slice(resumeAt), ...pending.slice(0, resumeAt)];
   const startedAt = Date.now();
   for (const op of ordered) {
+    if (options.shouldStop?.()) break;
     const remainingMs = deadlineMs - (Date.now() - startedAt);
-    if (state.outstanding.has(op.id) || remainingMs <= 0) {
+    if (state.outstanding.has(recoveryKey("mint", op.id)) || remainingMs <= 0) {
       unreachable++;
       continue;
     }
     state.after = op.id;
-    // Report on the refresh itself so a late result is still logged.
+    // Track refresh AND its late-result reporting mutations as one lifetime.
     const settled = source.ops.mint
       .refresh(op.id)
       .then(
         (result) => reportPendingMintRefresh(source, op, result, nowMs),
         (error) => reportPendingMintRefreshError(source, op, error),
-      )
-      .finally(() => state.outstanding.delete(op.id));
-    state.outstanding.set(op.id, settled);
+      );
+    trackRecovery(state.outstanding, recoveryKey("mint", op.id), settled);
     try {
       const outcome = await withTimeout(settled, Math.min(checkTimeoutMs, remainingMs));
       if (outcome === "unreachable") unreachable++;
@@ -1409,16 +1433,16 @@ async function reportPendingMintRefreshError(
  * still running after that fails against the closed database and is picked
  * up by startup recovery.
  */
-function startPendingMintSweep(source: PendingMintQuoteSource): () => Promise<void> {
+function startPendingMintSweep(source: PendingMintQuoteSource, outstanding: RecoveryWork): () => Promise<void> {
   let stopped = false;
   let timer: ReturnType<typeof setTimeout> | undefined;
   let inFlight: Promise<void> = Promise.resolve();
   let unreachableBefore = 0;
-  const state: PendingMintSweepState = { outstanding: new Map() };
+  const state: PendingMintSweepState = { outstanding };
 
   const tick = async () => {
     if (stopped) return;
-    inFlight = settlePendingMintQuotes(source, Date.now(), { state }).then(
+    inFlight = settlePendingMintQuotes(source, Date.now(), { state, shouldStop: () => stopped }).then(
       ({ unreachable }) => {
         // Report a mint becoming unreachable, or reachable again, once.
         if (unreachable > 0 && unreachableBefore === 0) {
@@ -1534,21 +1558,174 @@ interface RecoveryPhaseProgress {
   failedMintQuotes: number;
 }
 
+/**
+ * Coco keeps per-operation recovery private on its services; routstrd already
+ * reaches into the Manager the same way for `mintOperationService`. Send is
+ * the only family whose public `refresh()` cannot recover executing ops.
+ */
+function sendRecoveryServiceOf(coco: Manager): SendRecoveryService {
+  return (coco as unknown as { sendOperationService: SendRecoveryService })
+    .sendOperationService;
+}
+
+/** Local-only crash cleanup, run before the degraded gate opens. */
+export async function cleanupLocalRecoveryState(
+  coco: Manager,
+  repo: SqliteRepositories,
+): Promise<void> {
+  // Coco 1.0.1 implements these as local repository/proof operations only.
+  // Keep this version-sensitive bridge together with the send recovery bridge.
+  const services = coco as unknown as Record<string, {
+    recoverInitOperation?(op: unknown): Promise<void>;
+    cleanupOrphanedReservations?(): Promise<number>;
+  } | undefined>;
+  const families = [
+    ["send", repo.sendOperationRepository],
+    ["melt", repo.meltOperationRepository],
+    ["receive", repo.receiveOperationRepository],
+    ["mint", repo.mintOperationRepository],
+  ] as const;
+  // Fail closed the way reopenFailedMintOperation does: a coco bump that
+  // removes or renames these privates must stop recovery with a clear error
+  // before anything is written, not crash halfway through the loop with the
+  // cleanup half-applied.
+  for (const [kind] of families) {
+    if (typeof services[`${kind}OperationService`]?.recoverInitOperation !== "function") {
+      throw new Error(
+        `coco ${kind}OperationService.recoverInitOperation is unavailable; refusing local recovery cleanup`,
+      );
+    }
+  }
+  if (typeof services.sendOperationService?.cleanupOrphanedReservations !== "function") {
+    throw new Error(
+      "coco sendOperationService.cleanupOrphanedReservations is unavailable; refusing local recovery cleanup",
+    );
+  }
+  for (const [kind, repository] of families) {
+    for (const op of await repository.getByState("init")) {
+      await services[`${kind}OperationService`]!.recoverInitOperation!(op);
+    }
+  }
+  await services.sendOperationService!.cleanupOrphanedReservations!();
+}
+
+/**
+ * Gate for value-moving wallet operations while startup recovery runs.
+ *
+ * On degraded startup only, publishStuckMints opens the gate for callers
+ * whose target mint has no stuck operations after probing and local cleanup.
+ * A dead mint must not stall spends from a healthy one. On the happy path
+ * all callers wait until the global sweeps finish. Callers
+ * without a target mint, or whose mint has stuck operations, wait for the
+ * full sweep. fail() poisons every caller; reads are never gated.
+ */
+export interface RecoveryGate {
+  waitForRecovery(mintUrl?: string): Promise<void>;
+  publishStuckMints(mints: Set<string>): void;
+  complete(): void;
+  fail(error: string): void;
+}
+
+export function createRecoveryGate(): RecoveryGate {
+  let stuckMints: Set<string> | undefined;
+  let done = false;
+  let error: string | undefined;
+  let mintsResolve: (() => void) | undefined;
+  const mintsPromise = new Promise<void>((resolve) => {
+    mintsResolve = resolve;
+  });
+  let doneResolve: (() => void) | undefined;
+  const donePromise = new Promise<void>((resolve) => {
+    doneResolve = resolve;
+  });
+
+  return {
+    async waitForRecovery(mintUrl?: string): Promise<void> {
+      if (mintUrl) {
+        let normalized: string | undefined;
+        try {
+          normalized = normalizeMintUrl(mintUrl);
+        } catch {
+          // Unparseable URL falls back to the global gate.
+          normalized = undefined;
+        }
+        if (normalized) {
+          await mintsPromise;
+          if (!stuckMints?.has(normalized)) {
+            if (error) throw new Error(`Wallet is not ready: ${error}`);
+            return;
+          }
+        }
+      }
+      if (!done) await donePromise;
+      if (error) throw new Error(`Wallet is not ready: ${error}`);
+    },
+    publishStuckMints(mints: Set<string>): void {
+      if (stuckMints) return;
+      stuckMints = mints;
+      mintsResolve?.();
+    },
+    complete(): void {
+      done = true;
+      mintsResolve?.();
+      doneResolve?.();
+    },
+    fail(message: string): void {
+      done = true;
+      error = message;
+      mintsResolve?.();
+      doneResolve?.();
+    },
+  };
+}
+
 /**
  * Run the wallet recovery sweeps in order, reporting phase changes.
  *
- * Expired mint quotes are settled first: quotes their mint confirms as unpaid
+ * Probe first, then settle expired mint quotes: quotes confirmed as unpaid
  * are failed locally so `recoverPendingMintOperations()` skips them, while
  * paid/issued and unreachable-mint quotes stay pending for the sweep.
  */
-async function runWalletRecovery(
+export async function runWalletRecovery(
   coco: Manager,
   onProgress: (progress: RecoveryPhaseProgress) => void,
   receiveOperationIds?: string[],
+  onStuckMintsKnown?: (mints: Set<string>) => void,
+  options: { cleanupLocalState?: () => Promise<void>; fetchImpl?: typeof fetch; outstanding?: RecoveryWork; shouldStop?: () => boolean } = {},
 ): Promise<void> {
   surfacingRecoveryProgress = true;
   let failedMintQuotes = 0;
   try {
+    // Probe every mint that has stuck operations once, FIRST, so a dead mint
+    // costs a single short probe instead of taxing settlement's observation
+    // budget plus a network timeout per operation per sweep. Healthy-mint
+    // operations are recovered per op; dead-mint operations stay parked
+    // exactly as coco's own "will retry later" path would leave them.
+    onProgress({ phase: "Probing mints", failedMintQuotes });
+    const stuckOperations = await collectStuckOperations(coco.ops);
+    const unreachableMints = await probeMintReachability(
+      [...new Set(stuckOperations.map((op) => op.mintUrl))],
+      { fetchImpl: options.fetchImpl },
+    );
+    for (const mintUrl of unreachableMints) {
+      const count = stuckOperations.filter((op) => op.mintUrl === mintUrl).length;
+      startupProgress(
+        `Skipping recovery for unreachable mint: ${mintUrl} (${count} op${count === 1 ? "" : "s"})`,
+      );
+    }
+    const degraded = unreachableMints.size > 0;
+    if (degraded) {
+      // Local-only housekeeping must finish before any new operation is allowed.
+      await options.cleanupLocalState?.();
+      // Global sweeps enumerate fresh state and are unsafe beside live sends.
+      // Only the snapshot-based degraded path may open the per-mint gate.
+      onStuckMintsKnown?.(new Set(stuckOperations.map((op) => op.mintUrl)));
+    }
+
+    // Settlement runs after the gate opens and only spends its observation
+    // budget on mints the probe found reachable; dead-mint quotes stay
+    // pending untouched. It only reads and locally fails long-expired quotes,
+    // so it cannot conflict with live operations the gate just admitted.
     onProgress({ phase: "Settling expired mint quotes", failedMintQuotes });
     const settlement = await settleExpiredMintQuotes(
       {
@@ -1560,6 +1737,8 @@ async function runWalletRecovery(
         ).mintOperationService,
       },
       Date.now(),
+      undefined,
+      { unreachableMints, outstanding: options.outstanding, shouldStop: options.shouldStop },
     );
     failedMintQuotes = settlement.failed;
     if (settlement.leftForRecovery > 0 || settlement.unobserved > 0) {
@@ -1570,22 +1749,48 @@ async function runWalletRecovery(
       );
     }
     onProgress({ phase: "Settled expired mint quotes", failedMintQuotes });
+    const targeted = (kinds: Array<StuckOperation["kind"]>) =>
+      runTargetedRecovery(coco.ops, sendRecoveryServiceOf(coco), {
+        kinds,
+        outstanding: options.outstanding,
+        shouldStop: options.shouldStop,
+        stuckOperations,
+        unreachableMints,
+      });
 
+    // Happy path (every mint reachable) keeps coco's global sweeps: they also
+    // clean up init operations and orphaned proof reservations. Degraded
+    // startup runs that local housekeeping before opening its gate, and only
+    // drives the previously collected snapshot when a dead mint would
+    // otherwise tax every stuck op with a network timeout.
     onProgress({ phase: "Send recovery", failedMintQuotes });
-    await coco.ops.send.recovery.run();
+    if (!degraded) await coco.ops.send.recovery.run();
+    else await targeted(["send"]);
 
     onProgress({ phase: "Melt recovery", failedMintQuotes });
-    await coco.ops.melt.recovery.run();
+    if (!degraded) await coco.ops.melt.recovery.run();
+    else await targeted(["melt"]);
 
     onProgress({ phase: "Receive recovery", failedMintQuotes });
     if (receiveOperationIds) {
       // The pre-check already classified every executing receive by unique
       // input set. Recover only the conclusive retained operations; unresolved
       // groups stay untouched instead of falling back to Coco 1's expensive
-      // per-row sweep on this startup.
+      // per-row sweep on this startup. Operations at mints the probe found
+      // unreachable are skipped rather than costing their 15s timeout each.
+      const mintByOperation = new Map(
+        stuckOperations
+          .filter((op) => op.kind === "receive")
+          .map((op) => [op.id, op.mintUrl]),
+      );
       for (const operationId of receiveOperationIds) {
+        if (options.shouldStop?.()) break;
+        const mintUrl = mintByOperation.get(operationId);
+        if (mintUrl && unreachableMints.has(mintUrl)) continue;
         try {
-          await withTimeout(coco.ops.receive.refresh(operationId), 15_000);
+          const work = coco.ops.receive.refresh(operationId);
+          if (options.outstanding) trackRecovery(options.outstanding, recoveryKey("receive", operationId), work);
+          await withTimeout(work, 15_000);
         } catch (error) {
           logger.warn("Targeted receive recovery did not complete", {
             operationId,
@@ -1593,12 +1798,19 @@ async function runWalletRecovery(
           });
         }
       }
-    } else {
+    } else if (!degraded) {
       await coco.ops.receive.recovery.run();
+    } else {
+      await targeted(["receive"]);
     }
 
     onProgress({ phase: "Mint recovery", failedMintQuotes });
-    await coco.recoverPendingMintOperations();
+    // A settlement wait may have timed out while an unlocked observation
+    // still runs. Never let a fresh global mint sweep observe it again.
+    if (!degraded && ![...(options.outstanding?.keys() ?? [])].some(key => key.startsWith("mint:"))) {
+      await coco.recoverPendingMintOperations();
+    }
+    else await targeted(["mint"]);
 
     onProgress({ phase: "done", failedMintQuotes });
   } finally {
@@ -1661,6 +1873,10 @@ export async function createCocoClient(
   const recoveryPromise = new Promise<void>((resolve) => {
     recoveryResolve = resolve;
   });
+  const recoveryGate = createRecoveryGate();
+  let disposed = false;
+  const enqueueRecovery = createRunQueue();
+  const recoveryOutstanding: RecoveryWork = new Map();
 
   try {
     startupProgress("Opening Cashu wallet database...");
@@ -1858,11 +2074,14 @@ export async function createCocoClient(
         }
       },
       receiveRecoveryOperationIds,
+      (mints) => recoveryGate.publishStuckMints(mints),
+      { cleanupLocalState: () => cleanupLocalRecoveryState(coco!, repo), outstanding: recoveryOutstanding, shouldStop: () => disposed },
     )
       .then(async () => {
         await syncReceiveReservations();
         recoveryDone = true;
         recoveryPhase = "done";
+        recoveryGate.complete();
         recoveryResolve?.();
         startupProgress("Wallet recovery complete.");
         stopPendingMintSweep = startPendingMintSweep({
@@ -1871,12 +2090,13 @@ export async function createCocoClient(
           mintOperationService: (
             coco as unknown as { mintOperationService: MintOperationServiceCleanup }
           ).mintOperationService,
-        });
+        }, recoveryOutstanding);
       })
       .catch((error) => {
         recoveryDone = true;
         recoveryPhase = "error";
         recoveryError = error instanceof Error ? error.message : String(error);
+        recoveryGate.fail(recoveryError);
         recoveryResolve?.();
         startupProgress(`Wallet recovery failed: ${recoveryError}`);
       });
@@ -1899,23 +2119,33 @@ export async function createCocoClient(
     return api;
   };
 
-  let disposed = false;
-  // Explicit recovery runs are serialized, and finalize work that outlives its
-  // timeout stays in the map so a retry waits for it.
-  const enqueueRecovery = createRunQueue();
-  const recoveryOutstanding = new Map<string, Promise<unknown>>();
+  const assertOpen = () => { if (disposed) throw new Error("Wallet is shutting down"); };
 
-  /**
-   * Block a value-moving operation until background recovery has settled.
-   * Reads stay ungated so the daemon can report balances/status immediately.
-   */
-  const waitForRecovery = async (): Promise<void> => {
-    if (!recoveryDone) await recoveryPromise;
-    if (recoveryError) {
-      throw new Error(`Wallet is not ready: ${recoveryError}`);
-    }
+  // Block a value-moving operation until background recovery has settled for
+  // its target mint (see createRecoveryGate). Reads stay ungated so the
+  // daemon can report balances/status immediately.
+  const waitForRecovery = async (mintUrl?: string): Promise<void> => {
+    assertOpen();
+    await recoveryGate.waitForRecovery(mintUrl);
+    assertOpen();
   };
 
+  const disposeRecovery = createRecoveryDisposer(
+    () => { disposed = true; },
+    async () => {
+      await recoveryPromise;
+      await stopPendingMintSweep?.();
+      await enqueueRecovery.drain();
+      await drainRecoveryWork(recoveryOutstanding);
+    },
+    async () => {
+      await coco.dispose();
+      database.close();
+      releaseLegacyPidClaim();
+      releaseWalletPidClaim();
+    },
+  );
+
   return {
     async ping(): Promise<boolean> {
       try {
@@ -2094,13 +2324,13 @@ export async function createCocoClient(
     },
 
     async receiveBolt11(amount: number, mintUrl?: string) {
-      await waitForRecovery();
       const targetMint = mintUrl
         ? normalizeMintUrl(mintUrl)
         : walletConfig.defaultMintUrl;
       if (!targetMint) {
         throw new Error("No trusted mint available for Lightning invoice");
       }
+      await waitForRecovery(targetMint);
       const op = await coco.ops.mint.prepare({
         mintUrl: targetMint,
         amount,
@@ -2126,13 +2356,13 @@ export async function createCocoClient(
     },
 
     async sendCashu(amount: number, mintUrl?: string): Promise<string> {
-      await waitForRecovery();
       const targetMint = mintUrl
         ? normalizeMintUrl(mintUrl)
         : walletConfig.defaultMintUrl;
       if (!targetMint) {
         throw new Error("No trusted mint available for sending");
       }
+      await waitForRecovery(targetMint);
       const prepared = await coco.ops.send.prepare({
         mintUrl: targetMint,
         amount,
@@ -2142,13 +2372,13 @@ export async function createCocoClient(
     },
 
     async sendBolt11(invoice: string, mintUrl?: string): Promise<string> {
-      await waitForRecovery();
       const targetMint = mintUrl
         ? normalizeMintUrl(mintUrl)
         : walletConfig.defaultMintUrl;
       if (!targetMint) {
         throw new Error("No trusted mint available for Lightning payment");
       }
+      await waitForRecovery(targetMint);
       const prepared = await coco.ops.melt.prepare({
         mintUrl: targetMint,
         method: "bolt11",
@@ -2192,22 +2422,10 @@ export async function createCocoClient(
     },
 
     async dispose(): Promise<void> {
-      if (disposed) return;
-      disposed = true;
-      try {
-        // Let any in-flight recovery settle before closing the database from
-        // underneath it. The recovery promise resolves on success or failure.
-        await recoveryPromise;
-        await stopPendingMintSweep?.();
-        await coco.dispose();
-      } finally {
-        try {
-          database.close();
-        } finally {
-          releaseLegacyPidClaim();
-          releaseWalletPidClaim();
-        }
-      }
+      await disposeRecovery().catch(error => {
+        logger.warn("Wallet shutdown incomplete; database and ownership retained until recovery settles");
+        throw error;
+      });
     },
 
     async getHistory(offset?: number, limit?: number): Promise<HistoryEntry[]> {
@@ -2402,18 +2620,45 @@ export async function createCocoClient(
       // Serialize explicit recovery: two concurrent requests must not both
       // snapshot the same failed operation, and a retry must not start
       // underneath a finalize that outlived its timeout.
-      return enqueueRecovery(() =>
-        runMintQuoteRecovery(
+      return enqueueRecovery(() => {
+        assertOpen();
+        return runMintQuoteRecovery(
           {
             ops: coco.ops as unknown as MintQuoteRecoverySource["ops"],
             mintOperationService: service,
             reopenFailedOperation: (operationId) =>
               reopenFailedMintOperation(service, operationId),
           },
-          { ...options, outstanding: recoveryOutstanding },
+          { ...options, outstanding: recoveryOutstanding, shouldStop: () => disposed },
           onProgress,
-        ),
-      );
+        );
+      });
+    },
+
+    async recoverStuckOperations() {
+      await waitForRecovery();
+      // Serialized against explicit mint-quote recovery (and itself) through
+      // the same queue and lifetime tracker, so timed-out passes cannot retry the same
+      // operation. Receive stays startup-only: recovering competing receives
+      // safely requires the startup dedup classification (receive-dedup.ts).
+      // Operations a live execute holds come back as busy via coco's
+      // fail-fast operation lock, never driven underneath it.
+      const result = await enqueueRecovery(() => {
+        assertOpen();
+        return runTargetedRecovery(coco!.ops, sendRecoveryServiceOf(coco!), {
+          kinds: ["send", "melt", "mint"],
+          outstanding: recoveryOutstanding,
+          shouldStop: () => disposed,
+        });
+      });
+      return {
+        timedOut: result.timedOut,
+        attempted: result.attempted,
+        busy: result.busy,
+        skipped: result.skipped,
+        failed: result.failed,
+        skippedMints: Object.fromEntries(result.skippedMints),
+      };
     },
   };
 }

src/daemon/wallet/cocod-client.ts

@@ -159,6 +159,22 @@ export interface WalletMintQuoteRecoveryResult {
   errors: Array<{ operationId: string; error: string }>;
 }
 
+/** Summary of a stuck-operation (send/melt/mint) recovery run. */
+export interface WalletStuckOperationRecoveryResult {
+  /** Timed-out waits; the underlying operation remains tracked. */
+  timedOut: number;
+  /** Operations for which recovery was attempted (not necessarily completed). */
+  attempted: number;
+  /** Locked operations or unfinished work from another pass; retry later. */
+  busy: number;
+  /** Operations skipped for unreachable mints, shutdown, or pass budget exhaustion. */
+  skipped: number;
+  /** Operations at reachable mints whose recovery still failed. */
+  failed: number;
+  /** Unreachable mint URL -> number of operations skipped there. */
+  skippedMints: Record<string, number>;
+}
+
 export interface CocodClient {
   ping(): Promise<boolean>;
   getStatus(): Promise<CocodState>;
@@ -201,6 +217,12 @@ export interface CocodClient {
     options?: WalletMintQuoteRecoveryOptions,
     onProgress?: (message: string) => void,
   ): Promise<WalletMintQuoteRecoveryResult>;
+  /**
+   * Recover stuck send/melt/mint operations whose mints answer a
+   * reachability probe. Operations a live execute holds are reported busy,
+   * never driven. Receive stays startup-only (receive dedup classification).
+   */
+  recoverStuckOperations?(): Promise<WalletStuckOperationRecoveryResult>;
   /** Report background wallet recovery progress, when the wallet supports it. */
   getRecoveryProgress?(): Promise<WalletRecoveryProgress>;
 }

src/daemon/wallet/mint-quote-recovery.fake-mint.test.ts

@@ -363,7 +363,7 @@ describe("PAID mint quote recovery with a real Manager and mint", () => {
       outstanding,
     })) as unknown as Record<string, number>;
     expect(first).toMatchObject({ retryable: 1, recovered: 0 });
-    expect(outstanding.has(op.id as string)).toBe(true);
+    expect(outstanding.has(`mint:${op.id}`)).toBe(true);
 
     const second = (await runMintQuoteRecovery(booted.source() as never, {
       outstanding,

src/daemon/wallet/recovery-integration.test.ts

@@ -0,0 +1,257 @@
+import { expect, it, spyOn } from "bun:test";
+import { Manager } from "@cashu/coco-core";
+import { SqliteRepositories } from "@cashu/coco-sqlite-bun";
+import { Database } from "bun:sqlite";
+import { runTargetedRecovery, type SendRecoveryService } from "./recovery-probe";
+
+import {
+  cleanupLocalRecoveryState,
+  createRecoveryGate,
+  runWalletRecovery,
+} from "./coco-client";
+
+const MINT = "https://mint.example.com";
+const OP = "op-live";
+
+it("targeted recovery leaves a send that execute() holds alone", async () => {
+  const repo = new SqliteRepositories({ database: new Database(":memory:") });
+  await repo.init();
+  const coco = new Manager(repo, async () => new Uint8Array(64));
+  const internals = coco as unknown as {
+    sendOperationService: SendRecoveryService;
+    walletService: { getWalletWithActiveKeysetId: (m: string) => Promise<unknown> };
+  };
+
+  let swapStarted!: () => void;
+  const started = new Promise<void>((r) => (swapStarted = r));
+  let finishSwap!: (v: { send: unknown[]; keep: unknown[] }) => void;
+  const swap = new Promise<{ send: unknown[]; keep: unknown[] }>((r) => (finishSwap = r));
+  internals.walletService.getWalletWithActiveKeysetId = async () => ({
+    wallet: {
+      unit: "sat",
+      send: async () => (swapStarted(), swap),
+      checkProofsStates: async () => [{ state: "UNSPENT" }],
+      getFeesForProofs: () => 0,
+    },
+  });
+
+  await repo.proofRepository.saveProofs(MINT, [
+    { id: "00aa", amount: 8, secret: "in-1", C: "02aa", mintUrl: MINT, state: "ready" } as never,
+  ]);
+  await repo.proofRepository.reserveProofs(MINT, ["in-1"], OP);
+  await repo.sendOperationRepository.create({
+    id: OP, mintUrl: MINT, amount: 8, state: "prepared", method: "default", methodData: {},
+    createdAt: Date.now(), updatedAt: Date.now(), needsSwap: true, fee: 0, inputAmount: 8,
+    inputProofSecrets: ["in-1"],
+    outputData: {
+      keep: [],
+      send: [{
+        blindedMessage: { amount: 8, id: "00aa", B_: "02" + "11".repeat(32) },
+        blindingFactor: "01",
+        secret: Buffer.from("out-1").toString("hex"),
+      }],
+    },
+  } as never);
+
+  const live = coco.ops.send.execute(OP);
+  await started; // swap is at the mint
+  const result = await runTargetedRecovery(coco.ops, internals.sendOperationService, {
+    kinds: ["send"],
+    fetchImpl: (async () => new Response("{}")) as unknown as unknown as typeof fetch,
+  });
+  // On 8005aeb this is "rolled_back" and "in-1" is no longer reserved.
+  expect((await coco.ops.send.get(OP))?.state).toBe("executing");
+  // The live execute holds coco's per-operation lock: busy, never driven.
+  expect(result.attempted).toBe(0);
+  expect(result.busy).toBe(1);
+
+  finishSwap({ send: [{ id: "00aa", amount: 8, secret: "out-1", C: "02bb" }], keep: [] });
+  await live;
+  expect((await coco.ops.send.get(OP))?.state).toBe("pending");
+});
+
+
+it("keeps healthy-mint callers gated while happy-path global recovery runs", async () => {
+  const db = new Database(":memory:");
+  const repo = new SqliteRepositories({ database: db });
+  await repo.init();
+  const coco = new Manager(repo, async () => new Uint8Array(64));
+  const gate = createRecoveryGate();
+  let enter!: () => void;
+  const entered = new Promise<void>((resolve) => { enter = resolve; });
+  let release!: () => void;
+  const barrier = new Promise<void>((resolve) => { release = resolve; });
+  const spies = [
+    spyOn(coco.ops.send.recovery, "run").mockImplementation(async () => { enter(); await barrier; }),
+    spyOn(coco.ops.melt.recovery, "run").mockResolvedValue(undefined),
+    spyOn(coco.ops.receive.recovery, "run").mockResolvedValue(undefined),
+    spyOn(coco, "recoverPendingMintOperations").mockResolvedValue(undefined),
+  ];
+  try {
+    const recovery = runWalletRecovery(coco, () => {}, undefined,
+      (mints) => gate.publishStuckMints(mints),
+      { fetchImpl: (async () => new Response("{}")) as unknown as typeof fetch },
+    ).then(() => gate.complete());
+    await entered;
+    let released = false;
+    const waiting = gate.waitForRecovery(MINT).then(() => { released = true; });
+    await new Promise((resolve) => setTimeout(resolve, 10));
+    expect(released).toBe(false);
+    release();
+    await recovery;
+    await waiting;
+    expect(released).toBe(true);
+  } finally {
+    release();
+    for (const spy of spies) spy.mockRestore();
+    db.close();
+  }
+});
+
+it("degraded startup finishes local housekeeping before opening the per-mint gate", async () => {
+  const db = new Database(":memory:");
+  const repo = new SqliteRepositories({ database: db });
+  await repo.init();
+  const coco = new Manager(repo, async () => new Uint8Array(64));
+  await repo.sendOperationRepository.create({
+    id: "stuck", mintUrl: MINT, amount: 8, state: "rolling_back",
+    method: "default", methodData: {}, createdAt: Date.now(), updatedAt: Date.now(),
+  } as never);
+  const gate = createRecoveryGate();
+  const events: string[] = [];
+  const sweep = spyOn(coco.ops.send.recovery, "run");
+  try {
+    const recovery = runWalletRecovery(coco, () => {}, [], (mints) => {
+      events.push("gate");
+      gate.publishStuckMints(mints);
+    }, {
+      fetchImpl: (async () => { throw new Error("offline"); }) as unknown as unknown as typeof fetch,
+      cleanupLocalState: async () => { events.push("cleanup"); },
+    }).then(() => gate.complete());
+    await gate.waitForRecovery("https://healthy.example.com");
+    expect(events).toEqual(["cleanup", "gate"]);
+    await recovery;
+    expect(sweep).not.toHaveBeenCalled();
+    expect((await coco.ops.send.get("stuck"))?.state).toBe("rolling_back");
+  } finally {
+    sweep.mockRestore();
+    db.close();
+  }
+});
+
+it("degraded startup probes before settlement and never asks a dead mint", async () => {
+  const db = new Database(":memory:");
+  const repo = new SqliteRepositories({ database: db });
+  await repo.init();
+  const coco = new Manager(repo, async () => new Uint8Array(64));
+  const expiredRow = (id: string, mintUrl: string) => ({
+    id, mintUrl, quoteId: `q-${id}`, state: "pending",
+    createdAt: 1_000, updatedAt: 2_000, method: "bolt11",
+    methodData: { method: "bolt11", data: {} }, amount: 100, unit: "sat",
+    request: "lnbc1example", expiry: 1_000_000, // epoch seconds, long past
+  });
+  await repo.mintOperationRepository.create(expiredRow("dead-quote", "https://dead.example.com") as never);
+  await repo.mintOperationRepository.create(expiredRow("live-quote", "https://live.example.com") as never);
+  const internals = coco as unknown as {
+    mintOperationService: {
+      observePendingOperation(id: string): Promise<{ category: "waiting" }>;
+      failPendingOperation(op: unknown, failure: unknown): Promise<unknown>;
+    };
+  };
+  const events: string[] = [];
+  const observe = spyOn(internals.mintOperationService, "observePendingOperation")
+    .mockImplementation(async (id: string) => {
+      events.push(`observe:${id}`);
+      return { category: "waiting" };
+    });
+  // Fail the row for real (as the production service would) so the targeted
+  // mint pass sees state "failed" and refresh() returns without re-observing.
+  const fail = spyOn(internals.mintOperationService, "failPendingOperation")
+    .mockImplementation(async (op: unknown) => {
+      const id = (op as { id: string }).id;
+      const row = await repo.mintOperationRepository.getById(id);
+      if (row) await repo.mintOperationRepository.update({ ...row, state: "failed" } as never);
+      return {} as never;
+    });
+  const gate = createRecoveryGate();
+  try {
+    const recovery = runWalletRecovery(coco, () => {}, [], (mints) => {
+      events.push("gate");
+      gate.publishStuckMints(mints);
+    }, {
+      fetchImpl: (async (url: unknown) => {
+        if (String(url).includes("dead.example.com")) throw new Error("offline");
+        return new Response("{}");
+      }) as unknown as typeof fetch,
+      cleanupLocalState: async () => { events.push("cleanup"); },
+    }).then(() => gate.complete());
+    await recovery;
+    // The dead mint was probed, never asked to observe its quote; the gate
+    // opened before settlement spent anything on the live mint.
+    expect(events).toEqual(["cleanup", "gate", "observe:live-quote"]);
+    expect(observe).toHaveBeenCalledTimes(1);
+    expect(fail).toHaveBeenCalledTimes(1);
+  } finally {
+    observe.mockRestore();
+    fail.mockRestore();
+    db.close();
+  }
+});
+
+it("local housekeeping cleans init sends and orphaned reservations without network", async () => {
+  const db = new Database(":memory:");
+  const repo = new SqliteRepositories({ database: db });
+  await repo.init();
+  const coco = new Manager(repo, async () => new Uint8Array(64));
+  try {
+    await repo.sendOperationRepository.create({
+      id: "init-send", mintUrl: MINT, amount: 8, state: "init", method: "default",
+      methodData: {}, createdAt: Date.now(), updatedAt: Date.now(),
+    } as never);
+    for (const [id, repository] of [
+      ["init-melt", repo.meltOperationRepository],
+      ["init-receive", repo.receiveOperationRepository],
+      ["init-mint", repo.mintOperationRepository],
+    ] as const) {
+      await repository.create({
+        id, mintUrl: MINT, amount: 8, state: "init", method: "bolt11",
+        methodData: {}, inputProofs: [], createdAt: Date.now(), updatedAt: Date.now(),
+      } as never);
+    }
+    await repo.proofRepository.saveProofs(MINT, [
+      { id: "00aa", amount: 8, secret: "init-input", C: "02aa", mintUrl: MINT, state: "ready" },
+      { id: "00aa", amount: 8, secret: "orphan-input", C: "02aa", mintUrl: MINT, state: "ready" },
+    ] as never);
+    await repo.proofRepository.reserveProofs(MINT, ["init-input"], "init-send");
+    await repo.proofRepository.reserveProofs(MINT, ["orphan-input"], "missing-send");
+    await cleanupLocalRecoveryState(coco, repo);
+    expect(await coco.ops.send.get("init-send")).toBeNull();
+    expect(await repo.proofRepository.getReservedProofs()).toEqual([]);
+    expect(await repo.meltOperationRepository.getById("init-melt")).toBeNull();
+    expect(await repo.receiveOperationRepository.getById("init-receive")).toBeNull();
+    expect(await repo.mintOperationRepository.getById("init-mint")).toBeNull();
+  } finally {
+    db.close();
+  }
+});
+
+it("cleanup checks all private methods before making any local writes", async () => {
+  const db = new Database(":memory:");
+  const repo = new SqliteRepositories({ database: db });
+  await repo.init();
+  const coco = new Manager(repo, async () => new Uint8Array(64));
+  const internals = coco as unknown as { mintOperationService: { recoverInitOperation: unknown } };
+  const original = internals.mintOperationService.recoverInitOperation;
+  try {
+    await repo.sendOperationRepository.create({
+      id: "untouched-init", mintUrl: MINT, amount: 8, state: "init", method: "default",
+      methodData: {}, createdAt: Date.now(), updatedAt: Date.now(),
+    } as never);
+    internals.mintOperationService.recoverInitOperation = undefined;
+    await expect(cleanupLocalRecoveryState(coco, repo)).rejects.toThrow("mintOperationService.recoverInitOperation is unavailable");
+    expect((await repo.sendOperationRepository.getById("untouched-init"))?.state).toBe("init");
+  } finally {
+    internals.mintOperationService.recoverInitOperation = original;
+    db.close();
+  }
+});

src/daemon/wallet/recovery-probe.test.ts

@@ -0,0 +1,448 @@
+import { describe, expect, it } from "bun:test";
+import {
+  collectStuckOperations,
+  probeMintReachability,
+  runTargetedRecovery,
+  type StuckOperationSource,
+  type SendRecoveryService,
+} from "./recovery-probe";
+
+import { runMintQuoteRecovery, settlePendingMintQuotes, settleExpiredMintQuotes } from "./coco-client";
+
+interface FakeOp {
+  id: string;
+  mintUrl: string;
+  state: string;
+}
+
+function op(id: string, mintUrl: string, state = "pending"): FakeOp {
+  return { id, mintUrl, state };
+}
+
+interface FakeSource extends StuckOperationSource {
+  stuck: Record<"send" | "melt" | "receive" | "mint", FakeOp[]>;
+  refreshed: Record<"send" | "melt" | "receive" | "mint", string[]>;
+  failOnRefresh?: Set<string>;
+}
+
+function makeSource(
+  stuck: Partial<Record<"send" | "melt" | "receive" | "mint", FakeOp[]>>,
+): FakeSource {
+  const refreshed: FakeSource["refreshed"] = {
+    send: [],
+    melt: [],
+    receive: [],
+    mint: [],
+  };
+  const full = {
+    send: stuck.send ?? [],
+    melt: stuck.melt ?? [],
+    receive: stuck.receive ?? [],
+    mint: stuck.mint ?? [],
+  };
+  const family = (kind: keyof typeof full) => ({
+    diagnostics: { isLocked: () => false },
+    listInFlight: async () => full[kind] as never,
+    refresh: async (id: string) => {
+      refreshed[kind].push(id);
+      if (full[kind].some((o) => o.id === id && o.id.startsWith("boom"))) {
+        throw new Error("mint rejected the operation");
+      }
+      return full[kind].find((o) => o.id === id) as never;
+    },
+  });
+  return {
+    stuck: full,
+    refreshed,
+    send: {
+      ...family("send"),
+      get: async (id: string) =>
+        (full.send.find((o) => o.id === id) ?? null) as never,
+    } as FakeSource["send"],
+    melt: family("melt") as FakeSource["melt"],
+    receive: family("receive") as FakeSource["receive"],
+    mint: family("mint") as FakeSource["mint"],
+  };
+}
+
+function makeSendService(): SendRecoveryService & {
+  executingRecovered: string[];
+  /** Lock events in order, e.g. "acquire:op-1", "release:op-1". */
+  lockLog: string[];
+} {
+  const executingRecovered: string[] = [];
+  const lockLog: string[] = [];
+  return {
+    executingRecovered,
+    lockLog,
+    acquireOperationLock: async (id) => {
+      lockLog.push(`acquire:${id}`);
+      return () => {
+        lockLog.push(`release:${id}`);
+      };
+    },
+    recoverExecutingOperation: async (raw) => {
+      const id = (raw as FakeOp).id;
+      executingRecovered.push(id);
+      lockLog.push(`recover:${id}`);
+    },
+  };
+}
+
+/** fetch stub: mints whose URL contains "dead" hang/fail; others answer. */
+function makeFetch(deadPredicate: (url: string) => boolean) {
+  const calls: string[] = [];
+  const fetchImpl = (async (url: string | URL | Request) => {
+    const href = String(url);
+    calls.push(href);
+    if (deadPredicate(href)) throw new Error("connect ECONNREFUSED");
+    return new Response("{}", { status: 200 });
+  }) as unknown as typeof fetch;
+  return { calls, fetchImpl };
+}
+
+const isDeadUrl = (url: string) => url.includes("dead");
+
+const liveFetch = () => makeFetch(() => false);
+
+describe("collectStuckOperations", () => {
+  it("aggregates all four operation families with normalized mint URLs", async () => {
+    const source = makeSource({
+      send: [op("s1", "https://mint.example.com/")],
+      melt: [op("m1", "https://mint.example.com")],
+      receive: [op("r1", "https://other.example.com", "executing")],
+      mint: [op("q1", "https://mint.example.com")],
+    });
+    const stuck = await collectStuckOperations(source);
+    expect(stuck).toHaveLength(4);
+    expect(stuck.map((s) => s.kind).sort()).toEqual([
+      "melt",
+      "mint",
+      "receive",
+      "send",
+    ]);
+    // Trailing slash normalized so the same mint dedupes to one probe target.
+    const mintUrls = new Set(stuck.map((s) => s.mintUrl));
+    expect(mintUrls.size).toBe(2);
+  });
+
+  it("returns empty when nothing is stuck", async () => {
+    const stuck = await collectStuckOperations(makeSource({}));
+    expect(stuck).toEqual([]);
+  });
+});
+
+describe("probeMintReachability", () => {
+  it("marks only mints whose fetch fails as unreachable", async () => {
+    const deadFetch = makeFetch(isDeadUrl);
+    const unreachable = await probeMintReachability(
+      ["https://live.example.com", "https://dead.example.com"],
+      { fetchImpl: deadFetch.fetchImpl },
+    );
+    expect([...unreachable]).toEqual(["https://dead.example.com"]);
+    expect(deadFetch.calls).toHaveLength(2);
+    expect(deadFetch.calls.every((c) => c.endsWith("/v1/info"))).toBe(true);
+  });
+
+  it("treats HTTP error responses as reachable", async () => {
+    const fetchImpl = (async () =>
+      new Response("oops", { status: 500 })) as unknown as typeof fetch;
+    const unreachable = await probeMintReachability(["https://live.example.com"], {
+      fetchImpl,
+    });
+    expect(unreachable.size).toBe(0);
+  });
+});
+
+describe("runTargetedRecovery", () => {
+  it("does not probe when nothing is stuck", async () => {
+    const deadFetch = makeFetch(isDeadUrl);
+    const result = await runTargetedRecovery(makeSource({}), makeSendService(), {
+      fetchImpl: deadFetch.fetchImpl,
+    });
+    expect(result).toMatchObject({ attempted: 0, skipped: 0, failed: 0 });
+    expect(deadFetch.calls).toHaveLength(0);
+  });
+
+  it("skips operations at unreachable mints and recovers the rest", async () => {
+    const source = makeSource({
+      send: [op("s1", "https://dead.example.com"), op("s2", "https://live.example.com")],
+      melt: [op("m1", "https://dead.example.com"), op("m2", "https://live.example.com")],
+    });
+    const deadFetch = makeFetch(isDeadUrl);
+    const skippedMints: Array<[string, number]> = [];
+    const result = await runTargetedRecovery(source, makeSendService(), {
+      fetchImpl: deadFetch.fetchImpl,
+      onSkippedMint: (mintUrl, count) => skippedMints.push([mintUrl, count]),
+    });
+    expect(result.attempted).toBe(2);
+    expect(result.skipped).toBe(2);
+    expect(result.failed).toBe(0);
+    expect(skippedMints).toEqual([["https://dead.example.com", 2]]);
+    expect(result.skippedMints.get("https://dead.example.com")).toBe(2);
+    // Only live-mint operations were driven.
+    expect(source.refreshed.send).toEqual(["s2"]);
+    expect(source.refreshed.melt).toEqual(["m2"]);
+  });
+
+  it("recovers pending sends via refresh and executing sends via the service", async () => {
+    const source = makeSource({
+      send: [
+        op("pending-1", "https://live.example.com", "pending"),
+        op("exec-1", "https://live.example.com", "executing"),
+      ],
+    });
+    const sendService = makeSendService();
+    const result = await runTargetedRecovery(source, sendService, {
+      fetchImpl: liveFetch().fetchImpl,
+    });
+    expect(result.attempted).toBe(2);
+    expect(source.refreshed.send).toEqual(["pending-1"]);
+    expect(sendService.executingRecovered).toEqual(["exec-1"]);
+    // The executing send is driven under coco's per-operation lock.
+    expect(sendService.lockLog).toEqual([
+      "acquire:exec-1",
+      "recover:exec-1",
+      "release:exec-1",
+    ]);
+  });
+
+  it("re-reads state under the lock and skips a send that left executing", async () => {
+    const source = makeSource({
+      send: [op("s1", "https://live.example.com", "pending")],
+    });
+    const sendService = makeSendService();
+    // Snapshot taken while the op was still executing; it has since settled.
+    const result = await runTargetedRecovery(source, sendService, {
+      stuckOperations: [
+        {
+          kind: "send",
+          id: "s1",
+          mintUrl: "https://live.example.com",
+          state: "executing",
+          raw: op("s1", "https://live.example.com", "executing"),
+        },
+      ],
+      unreachableMints: new Set(),
+    });
+    expect(result.attempted).toBe(1);
+    expect(sendService.executingRecovered).toEqual([]);
+    // The lock is still acquired and released around the re-read.
+    expect(sendService.lockLog).toEqual(["acquire:s1", "release:s1"]);
+  });
+
+  it("counts a send as busy when a live execute wins the lock race", async () => {
+    const source = makeSource({
+      send: [op("s1", "https://live.example.com", "executing")],
+    });
+    const sendService = makeSendService();
+    sendService.acquireOperationLock = async () => {
+      const error = new Error("operation in progress");
+      error.name = "OperationInProgressError";
+      throw error;
+    };
+    const result = await runTargetedRecovery(source, sendService, {
+      fetchImpl: liveFetch().fetchImpl,
+    });
+    expect(result.busy).toBe(1);
+    expect(result.failed).toBe(0);
+    expect(sendService.executingRecovered).toEqual([]);
+  });
+
+  it("counts per-operation failures at reachable mints and continues", async () => {
+    const source = makeSource({
+      melt: [op("boom-1", "https://live.example.com"), op("m2", "https://live.example.com")],
+    });
+    const result = await runTargetedRecovery(source, makeSendService(), {
+      fetchImpl: liveFetch().fetchImpl,
+    });
+    expect(result.attempted).toBe(2);
+    expect(result.failed).toBe(1);
+    expect(source.refreshed.melt).toEqual(["boom-1", "m2"]);
+  });
+
+  it("respects the kinds filter and a pre-computed unreachable set", async () => {
+    const deadFetch = makeFetch(isDeadUrl);
+    const source = makeSource({
+      send: [op("s1", "https://dead.example.com")],
+      mint: [op("q1", "https://dead.example.com")],
+    });
+    const result = await runTargetedRecovery(source, makeSendService(), {
+      kinds: ["mint"],
+      unreachableMints: new Set(["https://dead.example.com"]),
+      fetchImpl: deadFetch.fetchImpl,
+    });
+    // No probe ran (pre-computed set) and the send op was not even counted.
+    expect(deadFetch.calls).toHaveLength(0);
+    expect(result.skipped).toBe(1);
+    expect(source.refreshed.send).toEqual([]);
+  });
+});
+
+
+it("preserves subpath mint URLs when probing", async () => {
+  const { fetchImpl, calls } = liveFetch();
+  await probeMintReachability(["https://mint.example.com/Bitcoin/"], { fetchImpl });
+  expect(calls).toEqual(["https://mint.example.com/Bitcoin/v1/info"]);
+});
+
+it("does not drive locked operations or count rolling-back sends as attempts", async () => {
+  const source = makeSource({ send: [
+    op("live", "https://mint.example.com", "executing"),
+    op("rollback", "https://mint.example.com", "rolling_back"),
+  ] });
+  source.send.diagnostics.isLocked = (id) => id === "live";
+  const service = makeSendService();
+  const result = await runTargetedRecovery(source, service, { fetchImpl: liveFetch().fetchImpl });
+  expect(result.attempted).toBe(0);
+  expect(result.busy).toBe(1);
+  expect(service.executingRecovered).toEqual([]);
+  expect(service.lockLog).toEqual([]);
+});
+
+it("classifies malformed persisted URLs as unreachable without fetching", async () => {
+  const { fetchImpl, calls } = liveFetch();
+  const unreachable = await probeMintReachability(["not-a-url"], { fetchImpl });
+  expect([...unreachable]).toEqual(["not-a-url"]);
+  expect(calls).toEqual([]);
+});
+
+it("bounds a hanging probe with an abort signal", async () => {
+  const fetchImpl = (async (_url: unknown, options: RequestInit) =>
+    new Promise<Response>((_resolve, reject) => {
+      options.signal!.addEventListener("abort", () => reject(options.signal!.reason), { once: true });
+    })) as unknown as typeof fetch;
+  const unreachable = await probeMintReachability(["https://slow.example.com"], {
+    fetchImpl, timeoutMs: 10,
+  });
+  expect([...unreachable]).toEqual(["https://slow.example.com"]);
+});
+
+
+it("shares timed-out mint observations across quote recovery, targeted recovery and the sweep", async () => {
+  const source = makeSource({ mint: [op("q1", "https://live.example.com")] });
+  let finish!: (value: { category: "waiting" }) => void;
+  const observation = new Promise<{ category: "waiting" }>(r => { finish = r; });
+  const outstanding = new Map<string, Promise<unknown>>();
+  const quoteSource = {
+    ops: { mint: {
+      listPending: async () => [{ ...source.stuck.mint[0]!, method: "bolt11" }],
+      get: async () => null,
+      finalize: async () => ({ state: "finalized" }),
+    } },
+    mintOperationService: { observePendingOperation: () => observation },
+    reopenFailedOperation: async () => false,
+  };
+  try {
+    await runMintQuoteRecovery(quoteSource as never, { outstanding, timeoutMs: 5 });
+    expect(outstanding.has("mint:q1")).toBe(true);
+    const result = await runTargetedRecovery(source, makeSendService(), {
+      outstanding, fetchImpl: liveFetch().fetchImpl,
+    });
+    expect(result.busy).toBe(1);
+    await settlePendingMintQuotes({
+      ops: { mint: { ...source.mint, listPending: async () => source.stuck.mint as never } },
+      wallet: { balances: { byMint: async () => ({}) } },
+      mintOperationService: { failPendingOperation: async () => ({}) },
+    } as never, Date.now(), { state: { outstanding } });
+    expect(source.refreshed.mint).toEqual([]);
+  } finally { finish({ category: "waiting" }); await observation; }
+});
+
+it("tracks a hung targeted mint so quote recovery skips it while unrelated operations proceed", async () => {
+  const source = makeSource({ mint: [op("q1", "https://live.example.com")], melt: [op("m1", "https://live.example.com")] });
+  let finish!: (value: never) => void;
+  source.mint.refresh = () => new Promise(r => { finish = r; });
+  const outstanding = new Map<string, Promise<unknown>>();
+  try {
+    const result = await runTargetedRecovery(source, makeSendService(), {
+      outstanding, timeoutMs: 5, fetchImpl: liveFetch().fetchImpl,
+    });
+    expect(result.timedOut).toBe(1);
+    expect(source.refreshed.melt).toEqual(["m1"]);
+    const quote = await runMintQuoteRecovery({
+      ops: { mint: { listPending: async () => [{ ...source.stuck.mint[0]!, method: "bolt11" }] } },
+    } as never, { outstanding });
+    expect(quote.busy).toBe(1);
+  } finally { finish({ state: "pending" } as never); }
+});
+
+it("retains executing-send lock until a timed-out underlying drive actually finishes", async () => {
+  const source = makeSource({ send: [op("s1", "https://live.example.com", "executing")] });
+  const service = makeSendService();
+  let finish!: () => void;
+  service.recoverExecutingOperation = () => new Promise<void>(r => { finish = r; });
+  const outstanding = new Map<string, Promise<unknown>>();
+  const result = await runTargetedRecovery(source, service, {
+    outstanding, timeoutMs: 5, fetchImpl: liveFetch().fetchImpl,
+  });
+  expect(result.timedOut).toBe(1);
+  expect(service.lockLog).toEqual(["acquire:s1"]);
+  const retry = await runTargetedRecovery(source, service, { outstanding, fetchImpl: liveFetch().fetchImpl });
+  expect(retry.busy).toBe(1);
+  const actual = outstanding.get("send:s1")!;
+  finish();
+  await actual;
+  expect(service.lockLog).toEqual(["acquire:s1", "release:s1"]);
+  expect(outstanding.size).toBe(0);
+});
+
+it("refuses missing executing-send internals without driving the operation", async () => {
+  const source = makeSource({ send: [op("s1", "https://live.example.com", "executing")] });
+  const result = await runTargetedRecovery(source, {} as SendRecoveryService, { fetchImpl: liveFetch().fetchImpl });
+  expect(result.failed).toBe(1);
+});
+
+
+it("startup settlement keeps its timed-out observation visible to manual recovery", async () => {
+  const source = makeSource({ mint: [op("q1", "https://live.example.com")] });
+  let finish!: (value: { category: "ready" }) => void;
+  const observation = new Promise<{ category: "ready" }>(r => { finish = r; });
+  const outstanding = new Map<string, Promise<unknown>>();
+  const settlement = await settleExpiredMintQuotes({
+    ops: { mint: { listPending: async () => [{ ...source.stuck.mint[0]!, expiry: 1, updatedAt: 1 }] } },
+    mintOperationService: { observePendingOperation: () => observation },
+  } as never, Date.now(), 5, { outstanding });
+  expect(settlement.unobserved).toBe(1);
+  const result = await runTargetedRecovery(source, makeSendService(), { outstanding, fetchImpl: liveFetch().fetchImpl });
+  expect(result.busy).toBe(1);
+  expect(source.refreshed.mint).toEqual([]);
+  const actual = outstanding.get("mint:q1")!;
+  finish({ category: "ready" });
+  await actual;
+  expect(outstanding.size).toBe(0);
+});
+
+it("deadline and shutdown leave remaining operations untouched", async () => {
+  const source = makeSource({ mint: [op("q1", "https://live.example.com")] });
+  const result = await runTargetedRecovery(source, makeSendService(), {
+    shouldStop: () => true, fetchImpl: liveFetch().fetchImpl,
+  });
+  expect(result.skipped).toBe(1);
+  expect(result.attempted).toBe(0);
+  expect(source.refreshed.mint).toEqual([]);
+});
+
+it("a pass budget bounds total drive waits, not just each operation", async () => {
+  const source = makeSource({ mint: [op("q1", "https://live.example.com"), op("q2", "https://live.example.com")] });
+  let finish!: (value: never) => void;
+  source.mint.refresh = () => new Promise(r => { finish = r; });
+  const outstanding = new Map<string, Promise<unknown>>();
+  const result = await runTargetedRecovery(source, makeSendService(), {
+    outstanding, timeoutMs: 100, deadlineMs: 10, fetchImpl: liveFetch().fetchImpl,
+  });
+  expect(result.timedOut).toBe(1);
+  expect(result.attempted).toBe(1);
+  expect(result.skipped).toBe(1);
+  const actual = outstanding.get("mint:q1")!;
+  finish({ state: "pending" } as never);
+  await actual;
+});
+
+it("unfinished work in another operation family does not block the same bare id", async () => {
+  const source = makeSource({ mint: [op("same-id", "https://live.example.com")] });
+  const outstanding = new Map<string, Promise<unknown>>([["send:same-id", new Promise(() => {})]]);
+  const result = await runTargetedRecovery(source, makeSendService(), { outstanding, fetchImpl: liveFetch().fetchImpl });
+  expect(result.busy).toBe(0);
+  expect(source.refreshed.mint).toEqual(["same-id"]);
+});

src/daemon/wallet/recovery-probe.ts

@@ -0,0 +1,312 @@
+/**
+ * Mint reachability probing and targeted (per-operation) wallet recovery.
+ *
+ * Coco's global recovery sweeps walk every non-terminal operation one at a
+ * time; each operation at an unreachable mint costs a full network timeout.
+ * This module probes every mint that has stuck operations once, in parallel,
+ * and drives recovery per operation only for mints that answer — so one dead
+ * mint costs a single short probe instead of N sequential timeouts, and its
+ * operations stay parked (exactly as coco's "will retry later" path leaves
+ * them) until an explicit mid-session recovery (see recoverStuckOperations
+ * in coco-client.ts) or a later startup finds the mint reachable.
+ */
+import { normalizeMintUrl, type Manager } from "@cashu/coco-core";
+import { recoveryKey, trackRecovery, waitForRecoveryWork, RecoveryWaitTimeout, type RecoveryWork } from "./recovery-work";
+import { logger } from "../../utils/logger";
+
+/** Short probe: a mint that cannot answer /v1/info in 2s slows every op. */
+export const MINT_PROBE_TIMEOUT_MS = 2_000;
+
+type OpsApi = Manager["ops"];
+
+/** Structural subset of the ops APIs used to enumerate and recover operations. */
+export interface StuckOperationSource {
+  send: Pick<OpsApi["send"], "listInFlight" | "refresh" | "diagnostics" | "get">;
+  melt: Pick<OpsApi["melt"], "listInFlight" | "refresh" | "diagnostics">;
+  receive: Pick<OpsApi["receive"], "listInFlight" | "refresh" | "diagnostics">;
+  mint: Pick<OpsApi["mint"], "listInFlight" | "refresh" | "diagnostics">;
+}
+
+export type StuckOperationKind = "send" | "melt" | "receive" | "mint";
+
+export interface StuckOperation {
+  kind: StuckOperationKind;
+  id: string;
+  /** Normalized mint URL. */
+  mintUrl: string;
+  state: string;
+  /** The operation object as returned by the API (needed by service-level recovery). */
+  raw: unknown;
+}
+
+/**
+ * The per-operation send recovery entry points coco keeps private. Send is the
+ * only operation family whose public `refresh()` does not cover `executing`
+ * operations; routstrd already reaches into coco internals the same way for
+ * `mintOperationService` (see coco-client.ts).
+ */
+export interface SendRecoveryService {
+  recoverExecutingOperation(op: unknown): Promise<void>;
+  /**
+   * coco's per-operation lock. It is fail-fast: acquiring an id a live
+   * execute/finalize/recover already holds throws OperationInProgressError
+   * instead of waiting. Holding it across the state re-read and the drive
+   * makes executing-send recovery atomic against a live execute — the same
+   * pattern reopenFailedMintOperation uses for mint operations
+   * (see coco-client.ts).
+   */
+  acquireOperationLock(operationId: string): Promise<() => void>;
+}
+
+export interface RecoveryRunResult {
+  /** Operations for which recovery was attempted (not necessarily completed). */
+  attempted: number;
+  /** Timed-out waits, also included in attempted; underlying work remains tracked. */
+  timedOut: number;
+  /** Locked operations or unfinished work from another pass; retry later. */
+  busy: number;
+  /** Operations skipped for unreachable mints, shutdown, or pass budget exhaustion. */
+  skipped: number;
+  /** Attempts that threw a non-busy, non-timeout error. */
+  failed: number;
+  /** Unreachable mint URL -> number of operations skipped there. */
+  skippedMints: Map<string, number>;
+}
+
+export interface TargetedRecoveryOptions {
+  outstanding?: RecoveryWork;
+  timeoutMs?: number;
+  deadlineMs?: number;
+  shouldStop?: () => boolean;
+  probeTimeoutMs?: number;
+  fetchImpl?: typeof fetch;
+  /** Operation families to recover. Defaults to all four. */
+  kinds?: StuckOperationKind[];
+  /** Pre-collected operations (e.g. from startup gating); default: enumerate now. */
+  stuckOperations?: StuckOperation[];
+  /** Pre-probed unreachable mints; default: probe now. */
+  unreachableMints?: Set<string>;
+  /** Called once per unreachable mint with the number of skipped operations. */
+  onSkippedMint?: (mintUrl: string, opCount: number) => void;
+}
+
+function asStuckOperations(
+  kind: StuckOperationKind,
+  ops: Array<{ id: string; mintUrl: string; state: string }>,
+): StuckOperation[] {
+  const stuck: StuckOperation[] = [];
+  for (const op of ops) {
+    try {
+      stuck.push({
+        kind,
+        id: op.id,
+        mintUrl: normalizeMintUrl(op.mintUrl),
+        state: op.state,
+        raw: op,
+      });
+    } catch {
+      // Preserve malformed persisted URLs; the probe will classify them as
+      // unreachable rather than attempting recovery against an invalid URL.
+      stuck.push({ kind, id: op.id, mintUrl: op.mintUrl, state: op.state, raw: op });
+    }
+  }
+  return stuck;
+}
+
+/**
+ * Enumerate every non-terminal operation across all four operation families.
+ * Returns [] quickly when nothing is stuck, which is the common case.
+ */
+export async function collectStuckOperations(
+  source: StuckOperationSource,
+): Promise<StuckOperation[]> {
+  const [sends, melts, receives, mints] = await Promise.all([
+    source.send.listInFlight(),
+    source.melt.listInFlight(),
+    source.receive.listInFlight(),
+    source.mint.listInFlight(),
+  ]);
+  return [
+    ...asStuckOperations("send", sends),
+    ...asStuckOperations("melt", melts),
+    ...asStuckOperations("receive", receives),
+    ...asStuckOperations("mint", mints),
+  ];
+}
+
+/**
+ * Probe each mint once, in parallel, and return the set of mint URLs that did
+ * not answer `GET /v1/info` in time. A mint that answers with an HTTP error is
+ * still "reachable" — its operations will fail with a real mint error instead
+ * of a network timeout, which is the information recovery needs.
+ */
+export async function probeMintReachability(
+  mintUrls: string[],
+  options: { timeoutMs?: number; fetchImpl?: typeof fetch } = {},
+): Promise<Set<string>> {
+  const fetcher = options.fetchImpl ?? fetch;
+  const timeoutMs = options.timeoutMs ?? MINT_PROBE_TIMEOUT_MS;
+  const unreachable = new Set<string>();
+
+  await Promise.all(
+    mintUrls.map(async (mintUrl) => {
+      try {
+        await fetcher(`${normalizeMintUrl(mintUrl)}/v1/info`, {
+          signal: AbortSignal.timeout(timeoutMs),
+        });
+      } catch (error) {
+        logger.debug("Mint did not answer recovery probe", {
+          mintUrl,
+          error: error instanceof Error ? error.message : String(error),
+        });
+        unreachable.add(mintUrl);
+      }
+    }),
+  );
+
+  return unreachable;
+}
+
+async function recoverStuckOperation(
+  source: StuckOperationSource,
+  sendService: SendRecoveryService,
+  op: StuckOperation,
+): Promise<void> {
+  switch (op.kind) {
+    case "send":
+      if (op.state === "pending") {
+        // Public API: actively re-checks the proofs with the mint.
+        await source.send.refresh(op.id);
+      } else if (op.state === "executing") {
+        // Fail closed: the lock and the drive are private coco internals
+        // (written against coco-core 1.0.1), so a coco bump that renames them
+        // must fail this operation loudly instead of corrupting it.
+        if (
+          typeof sendService.acquireOperationLock !== "function" ||
+          typeof sendService.recoverExecutingOperation !== "function"
+        ) {
+          throw new Error(
+            "coco sendOperationService recovery internals are unavailable; refusing to recover an executing send",
+          );
+        }
+        // recoverExecutingOperation takes no lock and does no state re-read,
+        // so take coco's fail-fast per-operation lock first (throws
+        // OperationInProgressError when a live execute holds the operation —
+        // the driver counts that as busy, not failed) and re-read the state
+        // under it: the snapshot op must still be executing before we drive.
+        const release = await sendService.acquireOperationLock(op.id);
+        try {
+          const latest = await source.send.get(op.id);
+          if (latest?.state === "executing") {
+            await sendService.recoverExecutingOperation(latest);
+          }
+        } finally {
+          release();
+        }
+      }
+      // prepared / rolling_back: the global sweep only warns; nothing to do.
+      return;
+    case "melt":
+      // refresh() covers both pending and executing melt operations.
+      await source.melt.refresh(op.id);
+      return;
+    case "receive":
+      // refresh() actively recovers executing receive operations.
+      await source.receive.refresh(op.id);
+      return;
+    case "mint":
+      // refresh() covers both pending and executing mint operations.
+      await source.mint.refresh(op.id);
+      return;
+  }
+}
+
+/**
+ * Recover every stuck operation whose mint answers a reachability probe,
+ * skipping operations at unreachable mints. Operations are recovered
+ * sequentially per mint (matching the global sweep's ordering guarantees);
+ * skipped operations are left untouched for a later pass, which is exactly
+ * what coco's own "Could not reach mint for recovery, will retry later" path
+ * does with them.
+ */
+export async function runTargetedRecovery(
+  source: StuckOperationSource,
+  sendService: SendRecoveryService,
+  options: TargetedRecoveryOptions = {},
+): Promise<RecoveryRunResult> {
+  const result: RecoveryRunResult = {
+    attempted: 0,
+    timedOut: 0,
+    busy: 0,
+    skipped: 0,
+    failed: 0,
+    skippedMints: new Map(),
+  };
+
+  const timeoutMs = options.timeoutMs ?? 15_000;
+  const deadlineMs = options.deadlineMs ?? 60_000;
+  if (![timeoutMs, deadlineMs].every(n => Number.isFinite(n) && n > 0)) {
+    throw new Error("Recovery budgets must be positive finite numbers");
+  }
+  const deadline = Date.now() + deadlineMs;
+  const outstanding = options.outstanding ?? new Map();
+  const kinds = options.kinds ?? ["send", "melt", "receive", "mint"];
+  const stuck = (options.stuckOperations ?? (await collectStuckOperations(source))).filter(
+    (op) => kinds.includes(op.kind),
+  );
+  if (stuck.length === 0) return result;
+
+  const unreachable =
+    options.unreachableMints ??
+    (await probeMintReachability([...new Set(stuck.map((op) => op.mintUrl))], {
+      timeoutMs: options.probeTimeoutMs,
+      fetchImpl: options.fetchImpl,
+    }));
+
+  for (const mintUrl of unreachable) {
+    const count = stuck.filter((op) => op.mintUrl === mintUrl).length;
+    result.skippedMints.set(mintUrl, count);
+    result.skipped += count;
+    options.onSkippedMint?.(mintUrl, count);
+  }
+
+  for (const op of stuck) {
+    if (unreachable.has(op.mintUrl)) continue;
+    if (options.shouldStop?.() || Date.now() >= deadline) { result.skipped++; continue; }
+    const key = recoveryKey(op.kind, op.id);
+    if (outstanding.has(key)) { result.busy++; continue; }
+    // Cheap pre-filter for live operations; the lock inside
+    // recoverStuckOperation is what actually makes the drive atomic.
+    if (source[op.kind].diagnostics.isLocked(op.id)) {
+      result.busy++;
+      continue;
+    }
+    if (op.kind === "send" && !["pending", "executing"].includes(op.state)) continue;
+    result.attempted++;
+    try {
+      const work = trackRecovery(outstanding, key, recoverStuckOperation(source, sendService, op));
+      await waitForRecoveryWork(work, Math.min(timeoutMs, Math.max(1, deadline - Date.now())));
+    } catch (error) {
+      if (error instanceof RecoveryWaitTimeout) { result.timedOut++; continue; }
+      // A live execute grabbed the operation between the isLocked pre-filter
+      // and the lock acquisition: busy, not failed — leave it for a later
+      // pass. Name-matched like runMintQuoteRecovery does, because the error
+      // crosses a package boundary.
+      if (error instanceof Error && error.name === "OperationInProgressError") {
+        result.busy++;
+        continue;
+      }
+      // Same semantics as coco's tryRecover*: leave the operation for the
+      // next pass. A reachable mint can still reject a specific operation.
+      result.failed++;
+      logger.warn("Targeted operation recovery did not complete", {
+        kind: op.kind,
+        operationId: op.id,
+        mintUrl: op.mintUrl,
+        error: error instanceof Error ? error.message : String(error),
+      });
+    }
+  }
+
+  return result;
+}

src/daemon/wallet/recovery-work.test.ts

@@ -0,0 +1,41 @@
+import { expect, it } from "bun:test";
+import { createRunQueue } from "./coco-client";
+import { createRecoveryDisposer, drainRecoveryWork, trackRecovery } from "./recovery-work";
+
+it("incomplete shutdown retains resources until late writes settle; retries join disposal", async () => {
+  const work = new Map<string, Promise<unknown>>();
+  let finish!: () => void;
+  const events: string[] = [];
+  trackRecovery(work, "mint:q", new Promise<void>(r => { finish = r; }).then(() => { events.push("write"); }));
+  const dispose = createRecoveryDisposer(() => {}, () => drainRecoveryWork(work), async () => { events.push("close"); }, 5);
+  await expect(dispose()).rejects.toThrow("Timed out");
+  expect(events).toEqual([]);
+  finish();
+  await dispose();
+  expect(events).toEqual(["write", "close"]);
+  await dispose();
+  expect(events).toEqual(["write", "close"]);
+});
+
+it("shutdown drains active queue work and queued callbacks reject before touching DB", async () => {
+  const queue = createRunQueue();
+  let disposed = false;
+  let finish!: () => void;
+  const events: string[] = [];
+  let started!: () => void;
+  const ready = new Promise<void>(r => { started = r; });
+  const active = queue(() => new Promise<void>(r => { finish = r; started(); }).then(() => { events.push("write"); }));
+  await ready;
+  const queued = queue(async () => {
+    if (disposed) throw new Error("Wallet is shutting down");
+    events.push("unexpected");
+  });
+  const rejected = queued.catch(error => error);
+  const dispose = createRecoveryDisposer(() => { disposed = true; }, () => queue.drain(), async () => { events.push("close"); }, 5);
+  await expect(dispose()).rejects.toThrow("Timed out");
+  finish();
+  await active;
+  expect((await rejected).message).toContain("shutting down");
+  await dispose();
+  expect(events).toEqual(["write", "close"]);
+});

src/daemon/wallet/recovery-work.ts

@@ -0,0 +1,45 @@
+/** A timed-out wait is not cancellation: retain work until it actually settles. */
+export type RecoveryWork = Map<string, Promise<unknown>>;
+export const recoveryKey = (kind: string, id: string): string => `${kind}:${id}`;
+
+export function trackRecovery<T>(work: RecoveryWork, key: string, promise: Promise<T>): Promise<T> {
+  work.set(key, promise);
+  const clear = () => { if (work.get(key) === promise) work.delete(key); };
+  void promise.then(clear, clear);
+  return promise;
+}
+
+export class RecoveryWaitTimeout extends Error {
+  constructor() { super("Timed out waiting for recovery; underlying work is still tracked"); }
+}
+
+export async function waitForRecoveryWork<T>(promise: Promise<T>, timeoutMs: number): Promise<T> {
+  let timer: ReturnType<typeof setTimeout> | undefined;
+  try {
+    return await Promise.race([promise, new Promise<never>((_, reject) => {
+      timer = setTimeout(() => reject(new RecoveryWaitTimeout()), timeoutMs);
+    })]);
+  } finally {
+    if (timer !== undefined) clearTimeout(timer);
+  }
+}
+
+/** Drain actual work, including work registered while an earlier task settles. */
+export async function drainRecoveryWork(work: RecoveryWork): Promise<void> {
+  while (work.size) await Promise.allSettled([...work.values()]);
+}
+
+/** Timeout reports incomplete disposal; actual cleanup continues safely. */
+export function createRecoveryDisposer(
+  quiesce: () => void,
+  settle: () => Promise<void>,
+  close: () => Promise<void>,
+  timeoutMs = 30_000,
+): () => Promise<void> {
+  let disposal: Promise<void> | undefined;
+  return async () => {
+    quiesce();
+    disposal ??= (async () => { await settle(); await close(); })();
+    await waitForRecoveryWork(disposal, timeoutMs);
+  };
+}