fix(recovery): continue twice-interrupted tasks and waiting follow-ups after a release swap (#21250)

## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work
> - When a run is cut off by a restart or a release swap, the
fresh-continuation fallback continues the task in a new run.
> - The continuation goes through normal admission, or through a
person's saved follow-up when one is waiting.
> - In release-swap waves, three gaps left tasks stopped with no run:
admission refused an isolated continuation of a task with no project, a
person's saved follow-up behind the replaced run was never promoted, and
the fallback relied on a saved follow-up that the cut-off run had
already read.
> - This pull request closes those three gaps.
> - The benefit is that a twice-interrupted task, and a task with a
waiting follow-up, continue by themselves.

## Linked Issues or Issue Description

Refs #21135, #21222

**What happened?**
1. A task with no project was interrupted twice. The second continuation
used a fresh isolated workspace, which clears the task's workspace
binding. The admission preflight still required that binding for an
isolated task with no project, so it refused the continuation as
`workspace_worktree_requires_project` about every 15 s. Each refusal
re-blocked the task and added a comment (329 to 428 skipped wakes per
wave). The runtime itself supports this case: it prepares a new
workspace (`create_new`).
2. The fallback continued a task with delivery `saved_input` (a person's
deferred follow-up is the next turn). The stranded-queue pass refused to
promote that follow-up, because the replaced native run's finalization
was `terminal_failure` (restart recovery ended it), not `committed`. The
release path would also have re-blocked the task with a reconciliation
incident. The follow-up never ran.
3. The fallback chose `saved_input` because a deferred user wake
existed, but the cut-off run had already taken that message into its
turn. Promotion settled the message as delivered, and no run followed.
The task showed "working".

**Expected behavior**
An interrupted task continues by itself in each case, with no person
action and no repeated comments.

**Steps to reproduce**
1. Run a release-swap wave with projectless tasks in isolated mode and
person follow-ups during runs.
2. Interrupt the same task twice, or interrupt a run that has a deferred
follow-up behind it.
3. See the task stay blocked or "working" with no run.

## What Changed

- Admission (`server/src/services/execution-workspace-policy.ts`,
`server/src/services/heartbeat/queue.ts`,
`server/src/services/fresh-continuation-fallback.ts`):
`isUnrunnableWorktreeCombo` accepts `freshIsolatedContinuation`. The
admission preflight sets it from the new
`isFreshIsolatedContinuationWake`, which returns at once (no query)
unless the wake context carries `freshContinuationIsolation`. This does
not touch the dispatch ranking.
- Follow-up promotion
(`server/src/modules/wake-queue/adapters/postgres.ts`): the
stranded-queue pass and the release path skip the native finalization
gate and the reconciliation incident for a run that the fallback
superseded (`agentRunSupersededByFreshContinuation`).
- Saved-input choice
(`server/src/services/fresh-continuation-fallback.ts`): a deferred user
wake counts as saved input only when it has no comment ids, or still has
a message that the agent has not received
(`undeliveredLegacyUserCommentIds`, called, not changed). Otherwise the
fallback records its own continuation.

- Review round 1: (a) whenever the old writer is not proven stopped
(`fresh_isolated_workspace`), the fallback records its own isolated
continuation first, also when the task had a binding; saved input
follows it on the new workspace. (b) Release skips the manual-admission
hold and the pending-cleanup check for a run the fallback replaced with
an isolated continuation (`fresh_isolated_workspace` or `remote_epoch`,
via `freshContinuationIsolatesRun`); other replacements still wait.
- Review round 2 (the isolated continuation must run before the waiting
follow-up): while the fallback's `fresh_isolated_workspace` continuation
is still pending delivery, release leaves the person's follow-up
deferred. Otherwise the follow-up would become the task's next run
without the isolation marker, and delivery would count it as the
continuation and skip the isolated run. The continuation's own release
promotes the follow-up behind it. `freshContinuationIsolatesRun` now
also reports `continuationPending`.

## Verification

Red-to-green tests (each failed before its fix and passes after it):
- `server/src/__tests__/execution-workspace-policy.test.ts`: admission
keeps a fresh isolated continuation of a projectless task runnable.
- `server/src/modules/wake-queue/adapters/postgres.test.ts`
"restart-ended native run": saved input behind a run the fallback
replaced is promoted.
- `server/src/services/fresh-continuation-fallback.test.ts` "relies on
saved input only while it still holds an undelivered message" (taken:
true gives `recorded`, taken: false gives `saved_input`).
- `postgres.test.ts` "promotes saved input behind an exhausted restart
hold ... (released, continuation pending)": the follow-up stays deferred
while the isolated continuation is pending, then is promoted when the
continuation's run is released. It was red before the fix (`expected
'promoted' to be 'released'`).
- Local, round 2: `cd server && npx vitest run
src/modules/wake-queue/adapters/postgres.test.ts
src/services/fresh-continuation-fallback.test.ts`: 2 files, 177 tests
passed.

Local command, on this branch based on current master (includes #21222
and #21234):
- `cd server && npx vitest run
src/services/fresh-continuation-fallback.test.ts
src/__tests__/execution-workspace-policy.test.ts
src/modules/wake-queue/adapters/postgres.test.ts`: 3 files, 221 tests
passed (re-run after rebasing on #21234).

Local release-swap cells:
- L1 (6 tasks, 3 swaps): 5 fresh isolated continuations delivered,
including a second one on the same task; 0
`workspace_worktree_requires_project` rows (before: 329 to 428 skipped
wakes per wave).
- R2 (6 tasks, 3 swaps): a task interrupted three times had its second
and third continuations delivered. The same cell found gap 3 (a
follow-up already read by the cut-off run), which the last commit fixes.
- Hour-long release-swap wave (w-20261112T233630-s7201) on the branch
that holds these fixes and the related recovery fixes: 11/11 tasks
finished hands-off (master: 9/11). The lost-continuation and
follow-up-promotion fixes were visibly working in it.

Review-round tests: "moves the task off the old writer's workspace and
runs the isolated continuation before saved input" (changed expectation:
`recorded` instead of `saved_input`; red on the previous head);
"promotes saved input behind an exhausted restart hold once a %s
continuation replaced the run" (isolated: promoted, red before;
`writer_ended`: held). 3 files, 224 tests passed.

Conflict note: the last commit was resolved onto current master. It no
longer imports `taskHistoryReadsForRun`, which belongs to a separate
opening-note change.

## Risks

- An isolated projectless continuation now passes admission. It runs in
a new workspace that the run prepares, so it never shares a workspace
with the cut-off run.
- A superseded run's follow-up is promoted without its native
finalization committed. The superseded run is fenced (its writes are
refused), so there is still one writer.
- No migration. No schema change. The normal admission path adds no
query.

## Model Used

- Claude Opus 5.5 (`claude-opus-5-5`), Claude Code agent with tool use
(shell, file edit, test runs).

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] I have updated relevant documentation to reflect my changes (no
documented behavior or command changes)
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
DottaandClaude Opus 5.5 authored and GitHub committed 2026-11-12 21:10:47 -06:00
1 parent a298ca7fbc
commit 6cb0dd0c6b
7 files changed
+200 -16

No files matched your search

@@ -154,6 +154,14 @@ describe("execution workspace policy helpers", () => {
).toEqual({ type: "git_worktree", existingBranch: "PAP-14754-run-redaction" });
});
it("admits a fresh isolated continuation of a projectless task whose binding the fallback cleared", () => {
const issue = { projectId: null, projectWorkspaceId: null, executionWorkspaceId: null,
executionWorkspacePreference: "reuse_existing" };
expect(isUnrunnableWorktreeCombo({ issue, resolvedMode: "isolated_workspace", resolvedStrategy: "git_worktree" })).toBe(true);
expect(isUnrunnableWorktreeCombo({ issue, resolvedMode: "isolated_workspace", resolvedStrategy: "git_worktree",
freshIsolatedContinuation: true })).toBe(false);
});
it("centralizes unrunnable isolated worktree detection", () => {
expect(
isUnrunnableWorktreeCombo({
@@ -2,9 +2,10 @@ import { resolveExternalTestDatabaseUrl } from "../../../__tests__/helpers/exter
import { DELIVERY_QUEUES } from "../../../services/delivery-work-notifications.js";
import { instanceSettingsService } from "../../../services/instance-settings.js";
import { randomBytes, randomUUID } from "node:crypto";
import { and, eq } from "drizzle-orm";
import { and, eq, sql } from "drizzle-orm";
import { afterAll, afterEach, beforeAll, describe, expect, it, vi } from "vitest";
import type { Db } from "@paperclipai/db";
import { createSystemRecovery } from "../../../services/system-recovery-state.js";
import {
activityLog,
agentWakeupRequests,
@@ -474,6 +475,85 @@ describeEmbeddedPostgres("wake-queue postgres adapter", () => {
expect((await release({ companyId, runId, now: new Date(), suppressImmediateRecovery: true, resumeStrandedQueue: true })).outcome.kind).toBe("promoted");
});
it.each([true, false])("promotes saved input behind a restart-ended native run only once the fresh continuation replaced it (replaced: %s)", async replaced => {
const companyId = await seedCompany(), agentId = await seedAgent({ companyId });
const issueId = await seedIssue({ companyId, status: "in_progress", assigneeAgentId: agentId });
const runId = await seedRun({ companyId, agentId, status: "failed", contextSnapshot: { issueId } });
await db.update(heartbeatRuns).set({ runtimeMode: "native", nativeIssueId: issueId, finishedAt: new Date(),
errorCode: "native_restart_recovery_blocked",
resultJson: replaced ? { freshContinuationSupersededAt: new Date().toISOString() } : {},
}).where(eq(heartbeatRuns.id, runId));
// Restart recovery ended the run: its finalization stays terminal_failure, never committed.
await db.insert(nativeRunFinalizations).values({ companyId, issueId, runId, phase: "terminal_failure",
recoveryState: "blocked", failureCode: "native_restart_recovery_blocked" });
const [comment] = await db.insert(issueComments).values({ companyId, issueId, authorType: "user", authorUserId: "responsible-user", body: "Follow-up during the run" }).returning();
const wakeId = await seedDeferredWake({ companyId, agentId, issueId, requestedByActorId: "responsible-user",
payload: { _paperclipWakeContext: { issueId, wakeReason: "issue_commented", wakeCommentIds: [comment.id] } },
});
const release = createReleaseIssueExecution({ issueLock: createPostgresWakeQueueAdapter(db, stubDeps),
recovery: { escalateStrandedAssignedIssue: async () => undefined, escalateStrandedRecoveryIssueInPlace: async () => undefined },
});
const result = await release({ companyId, runId, now: new Date(), suppressImmediateRecovery: true, resumeStrandedQueue: true });
if (replaced) {
expect(result.outcome.kind).toBe("promoted");
expect((await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.id, wakeId)))[0].status).not.toBe("deferred_issue_execution");
// No reconciliation incident re-blocks the task the fallback released.
expect((await db.select().from(issues).where(eq(issues.id, issueId)))[0].status).not.toBe("blocked");
} else {
expect(result.outcome.kind).toBe("released");
expect((await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.id, wakeId)))[0].status).toBe("deferred_issue_execution");
}
});
it.each([
["fresh_isolated_workspace", "promoted", "not_requested"], ["remote_epoch", "promoted", "not_requested"],
["writer_ended", "released", "not_requested"],
// The isolated continuation is still to be delivered: it runs first and binds the new workspace;
// promoting the follow-up now would make it the task's next run, without the isolation.
["fresh_isolated_workspace", "released", "pending"],
] as const)("promotes saved input behind an exhausted restart hold once a %s continuation replaced the run (%s, continuation %s)", async (isolation, expected, continuationDelivery) => {
// Restart recovery exhausted its retries (systemRecovery.state exhausted, no stop receipt), then the
// fallback replaced the run. An isolated continuation reuses none of the old executor's workspace,
// so the manual-admission hold no longer stands between the task and the person's follow-up.
const companyId = await seedCompany(), agentId = await seedAgent({ companyId });
const issueId = await seedIssue({ companyId, status: "in_progress", assigneeAgentId: agentId });
const runId = await seedRun({ companyId, agentId, status: "failed", contextSnapshot: { issueId } });
await db.update(heartbeatRuns).set({ runtimeMode: "native", nativeIssueId: issueId, finishedAt: new Date(),
errorCode: "native_restart_recovery_blocked", resultJson: { freshContinuationSupersededAt: new Date().toISOString() },
}).where(eq(heartbeatRuns.id, runId));
await db.insert(nativeRunFinalizations).values({ companyId, issueId, runId, phase: "terminal_failure",
recoveryState: "blocked", failureCode: "native_restart_recovery_blocked",
failureDetail: { systemRecovery: { ...createSystemRecovery(runId), state: "exhausted" }, replacementDenied: "fresh_continuation" } });
await db.insert(issueRecoveryActions).values({ companyId, sourceIssueId: issueId, kind: "active_run_watchdog",
status: "resolved", outcome: "restored", ownerType: "system", returnOwnerAgentId: agentId,
cause: "fresh_continuation", fingerprint: `fresh-continuation:${runId}`, nextAction: "Continuing.", resolvedAt: new Date(),
evidence: { runId, freshContinuation: { sourceRunId: runId, isolation,
delivery: continuationDelivery === "pending" ? "recorded" : "saved_input" }, continuationDelivery } });
const [comment] = await db.insert(issueComments).values({ companyId, issueId, authorType: "user", authorUserId: "responsible-user", body: "Follow-up during the run" }).returning();
const wakeId = await seedDeferredWake({ companyId, agentId, issueId, requestedByActorId: "responsible-user",
payload: { _paperclipWakeContext: { issueId, wakeReason: "issue_commented", wakeCommentIds: [comment.id] } },
});
const release = createReleaseIssueExecution({ issueLock: createPostgresWakeQueueAdapter(db, stubDeps),
recovery: { escalateStrandedAssignedIssue: async () => undefined, escalateStrandedRecoveryIssueInPlace: async () => undefined },
});
const result = await release({ companyId, runId, now: new Date(), suppressImmediateRecovery: true, resumeStrandedQueue: true });
expect(result.outcome.kind).toBe(expected);
const [wake] = await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.id, wakeId));
if (expected === "promoted") expect(wake.status).not.toBe("deferred_issue_execution");
else expect(wake.status).toBe("deferred_issue_execution");
if (continuationDelivery !== "pending") return;
// The isolated continuation is delivered and runs; its release promotes the follow-up behind it.
const continuationRunId = await seedRun({ companyId, agentId, status: "succeeded", contextSnapshot: { issueId } });
await db.update(heartbeatRuns).set({ finishedAt: new Date() }).where(eq(heartbeatRuns.id, continuationRunId));
await db.update(issueRecoveryActions).set({ evidence: sql`${issueRecoveryActions.evidence}
|| ${JSON.stringify({ continuationDelivery: "delivered", continuationRunId })}::jsonb` })
.where(eq(issueRecoveryActions.sourceIssueId, issueId));
expect((await release({ companyId, runId: continuationRunId, now: new Date(), suppressImmediateRecovery: true })).outcome.kind)
.toBe("promoted");
expect((await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.id, wakeId)))[0].status)
.not.toBe("deferred_issue_execution");
});
it.each([
"completed", "multiple_comments", "repeated_reference", "parent_reference", "mixed_issue_references", "mixed_foreign_references", "mixed_unknown_references", "no_comments", "missing_comment",
"human_comment", "other_run_comment", "foreign_comment", "other_issue_comment", "deleted_comment",
@@ -18,6 +18,8 @@ import { nativeManualAdmissionHoldsExecution } from "../../../services/native-ru
import { and, asc, eq, inArray, isNull, notInArray, or, sql } from "drizzle-orm";
import { readReleaseRun, type ReleaseRun } from "./release-run-read.js";
import { isGovernancePauseStop } from "../../../services/governance-pause-stop.js";
import { agentRunSupersededByFreshContinuation } from "../../../agent-run-cancellation.js";
import { freshContinuationIsolatesRun } from "../../../services/fresh-continuation-fallback.js";
import type { Db } from "@paperclipai/db";
import { deliveredQueuedCommentIds, EXTERNAL_CHAT_MESSAGE_WAKE_MUTATION, EXTERNAL_CHAT_MESSAGE_WAKE_REASON } from "../../../services/queued-comment-delivery.js";
import { extractIssueReferenceIdentifiers, isUuidLike } from "@paperclipai/shared";
@@ -1284,7 +1286,10 @@ export function createPostgresWakeQueueAdapter(db: Db, deps: WakeQueuePostgresAd
`,
);
if (input.resumeStrandedQueue && run.runtimeMode === "native") {
// A run the fresh-continuation fallback replaced is settled for good: its
// finalization may stay terminal_failure (restart recovery ended it), and
// the saved input queued behind it is now the task's next turn.
if (input.resumeStrandedQueue && run.runtimeMode === "native" && !agentRunSupersededByFreshContinuation(run)) {
const [finalized] = contextIssueId && run.nativeIssueId === contextIssueId
? await tx.select({ runId: nativeRunFinalizations.runId }).from(nativeRunFinalizations).where(and(
eq(nativeRunFinalizations.companyId, input.companyId), eq(nativeRunFinalizations.runId, run.id),
@@ -1293,10 +1298,16 @@ export function createPostgresWakeQueueAdapter(db: Db, deps: WakeQueuePostgresAd
: [];
if (!finalized) return { outcome: { kind: "released" }, postCommitEffects: [], run: runSnapshot };
}
if (await nativeManualAdmissionHoldsExecution(tx, { companyId: input.companyId, runId: run.id })) {
// These two holds protect the old executor's workspace and resources. A
// continuation in a new isolated workspace (or an abandoned remote epoch)
// reuses none of them, so they no longer hold the task's next turn; any
// other replacement still waits for them.
const isolatedReplacement = agentRunSupersededByFreshContinuation(run)
? await freshContinuationIsolatesRun(tx, { companyId: input.companyId, runId: run.id }) : false;
if (!isolatedReplacement && await nativeManualAdmissionHoldsExecution(tx, { companyId: input.companyId, runId: run.id })) {
return { outcome: { kind: "released" }, postCommitEffects: [], run: runSnapshot };
}
if (input.resumeStrandedQueue && await hasPendingRunCleanup(tx, input.companyId, run.id)) {
if (input.resumeStrandedQueue && !isolatedReplacement && await hasPendingRunCleanup(tx, input.companyId, run.id)) {
return { outcome: { kind: "released" }, postCommitEffects: [], run: runSnapshot };
}
@@ -1448,13 +1459,25 @@ export function createPostgresWakeQueueAdapter(db: Db, deps: WakeQueuePostgresAd
&& issueRow.status === "blocked" && issueRow.assigneeAgentId === run.agentId) {
return { outcome: { kind: "released" }, postCommitEffects: [], run: runSnapshot };
}
if (!handoffAssigneeAgentId && await recordNativeTerminalRecoveryIfNeeded(tx, run, issueRow, input.now)) {
// The fallback already decided this run's continuation; a reconciliation
// incident would re-block the task it just released.
if (!handoffAssigneeAgentId && !agentRunSupersededByFreshContinuation(run)
&& await recordNativeTerminalRecoveryIfNeeded(tx, run, issueRow, input.now)) {
return { outcome: { kind: "released" }, postCommitEffects: [], run: runSnapshot };
}
// A release must leave deferred messages intact while execution is held.
// The finishing conversation may still own its lease until finally cleanup.
// Allow only bounded retry planning in that case; admission stays gated.
// The fallback's isolated continuation runs before any saved input: it
// alone carries the marker that prepares the new workspace and binds it.
// Promoting the follow-up now would make it the task's next run without
// that isolation (and the delivery worker would then skip the
// continuation). The follow-up stays deferred and is promoted behind
// the continuation when that run is released.
if (isolatedReplacement && isolatedReplacement.continuationPending) {
return { outcome: { kind: "released" }, postCommitEffects: [], run: runSnapshot };
}
const executionBlocker = await getExecutionBlocker(tx, issueRow.companyId, issueRow.id);
const recoveryOnly = Boolean(executionBlocker &&
executionBlocker.cause === "execution_owner_active" && executionBlocker.runId === run.id &&
@@ -94,7 +94,11 @@ export function isUnrunnableWorktreeCombo(input: {
resolvedStrategy: string | null | undefined;
reusableExecutionWorkspaceAvailable?: boolean | null;
hasResolvablePriorSessionWorkspace?: boolean | null;
/** The wake continues a cut-off run in a new isolated workspace that the run
* prepares itself (`create_new`); it never needs the task's old binding. */
freshIsolatedContinuation?: boolean | null;
}): boolean {
if (input.freshIsolatedContinuation === true) return false;
if (input.resolvedMode !== "isolated_workspace" && input.resolvedMode !== "operator_branch") return false;
if (input.resolvedStrategy !== "git_worktree") return false;
if (input.issue.projectId || input.issue.projectWorkspaceId) return false;
@@ -3,7 +3,7 @@ import { and, eq, sql } from "drizzle-orm";
import { afterAll, afterEach, beforeAll, describe, expect, it, vi } from "vitest";
import {
activityLog, agentWakeupRequests, agents, approvals, authUsers, companies, createDb, environmentLeases, environments, executionWorkspaces, heartbeatRuns, issueApprovals, issueRecoveryActions, issueThreadInteractions,
issues, nativeRunFinalizations, toolInvocations,
issueComments, issues, nativeRunFinalizations, toolInvocations,
} from "@paperclipai/db";
import { getEmbeddedPostgresTestSupport, startEmbeddedPostgresTestDatabase } from "../__tests__/helpers/embedded-postgres.js";
import { deliverReconciledExecutions, settleUnrecoverableExecutions } from "./execution-recovery-resolution.js";
@@ -803,6 +803,22 @@ suite("fresh-continuation fallback", () => {
expect(decision.outcomeEvidence).not.toMatch(adminText);
}, 60_000);
it.each([true, false])("relies on saved input only while it still holds an undelivered message (already taken by the cut-off run: %s)", async taken => {
// R2 cell t003: the cut-off run had taken the follow-up into its turn; the fallback chose saved_input,
// promotion settled the message as delivered, and no run followed.
const f = await unfinishedRun({ runtime: "native", errorCode: "native_restart_recovery_blocked" });
const [comment] = await db.insert(issueComments).values({ companyId, issueId: f.issueId, authorType: "user",
authorUserId: userId, body: "Additional requirement: include FOLLOWUP-1" } as typeof issueComments.$inferInsert).returning();
if (taken) await db.update(heartbeatRuns).set({ contextSnapshot: { issueId: f.issueId, wakeCommentIds: [comment!.id] } })
.where(eq(heartbeatRuns.id, f.runId));
await db.insert(agentWakeupRequests).values({ companyId, agentId, source: "automation", triggerDetail: "system",
reason: "issue_execution_deferred", status: "deferred_issue_execution", requestedByActorType: "user", requestedByActorId: userId,
payload: { issueId: f.issueId, _paperclipWakeContext: { issueId: f.issueId, wakeReason: "issue_commented", wakeCommentIds: [comment!.id] } },
} as typeof agentWakeupRequests.$inferInsert);
expect(await ensureFreshContinuation(db, { companyId, runId: f.runId, trigger: "recovery_refused" }))
.toMatchObject({ kind: "continued", delivery: taken ? "recorded" : "saved_input" });
}, 60_000);
it.each([
{ title: "a pending approval", setup: "approval" as const, errorCode: "native_continuation_requires_reconciliation" },
{ title: "a credential the person has not shared", setup: "none" as const, errorCode: "configuration_incomplete" },
@@ -881,18 +897,20 @@ suite("fresh-continuation fallback", () => {
}, 60_000);
// The isolation decision is a property of the task's next run, whatever path starts it.
it("moves the task off the old writer's workspace for every next run, saved input included", async () => {
it("moves the task off the old writer's workspace and runs the isolated continuation before saved input", async () => {
const f = await unfinishedRun({ runtime: "legacy", errorCode: "controller_lease_lost", error: "Legacy controller lease lost" });
const [workspace] = await db.insert(executionWorkspaces).values({ companyId, sourceIssueId: f.issueId, mode: "isolated",
strategyType: "git_worktree", name: "Task workspace", cwd: "/tmp/old-writer" }).returning();
await db.update(issues).set({ executionWorkspaceId: workspace!.id, workspaceBindingRevision: 2 }).where(eq(issues.id, f.issueId));
await db.update(heartbeatRuns).set({ processPid: 4242, processHostHash: "replaced-server", processStartedAt: new Date(Date.now() - 600_000),
contextSnapshot: { issueId: f.issueId, executionWorkspaceId: workspace!.id } }).where(eq(heartbeatRuns.id, f.runId));
// A person's saved message is waiting: it continues the task itself.
// A person's saved message is waiting. It carries no isolation marker, so it cannot prepare the
// new workspace (a projectless task would be refused at admission once the binding is cleared):
// the isolated continuation runs first and the saved message follows it.
await db.insert(agentWakeupRequests).values({ companyId, agentId, source: "on_demand", status: "deferred_issue_execution",
requestedByActorType: "user", requestedByActorId: userId, payload: { issueId: f.issueId } });
expect(await ensureFreshContinuation(db, { companyId, runId: f.runId, trigger: "recovery_refused" }))
.toMatchObject({ kind: "continued", isolation: "fresh_isolated_workspace", delivery: "saved_input" });
.toMatchObject({ kind: "continued", isolation: "fresh_isolated_workspace", delivery: "recorded" });
// No delivery path can reach the old directory: the task is no longer bound to it.
expect(await issue(f.issueId)).toMatchObject({ executionWorkspaceId: null, workspaceBindingRevision: 3 });
const [kept] = await db.select().from(executionWorkspaces).where(eq(executionWorkspaces.id, workspace!.id));
@@ -8,6 +8,8 @@ import {
} from "@paperclipai/shared";
import { logger } from "../middleware/logger.js";
import { logActivity } from "./activity-log.js";
import { undeliveredLegacyUserCommentIds } from "./explicit-native-continuation.js";
import { queuedCommentIdsFromWakePayload } from "./issue-queued-comment-queue.js";
import { FRESH_CONTINUATION_SUPERSEDED_KEY } from "../agent-run-cancellation.js";
import { isAcknowledgedNativeReassignmentStop, isAcknowledgedNativeStop } from "./acknowledged-native-stop.js";
import { adapterExecutionControls } from "./adapter-execution-control.js";
@@ -305,6 +307,40 @@ export async function readFreshContinuationReadiness(db: Db, run: Pick<typeof he
return { mode: "fresh_isolated_workspace" as const, retryIntentId: recordId, sourceRunId, retainedSourceIds: [] as string[] };
}
/**
* True when a wake continues a cut-off run in a new isolated workspace. The
* fallback cleared the task's workspace binding on purpose, and the run
* prepares a new workspace itself, so admission must not refuse the wake for
* lacking that binding (it would refuse it forever).
*/
export async function isFreshIsolatedContinuationWake(db: Db, input: { companyId: string; issueId: string;
contextSnapshot: Record<string, unknown> | null | undefined }) {
if (typeof object(input.contextSnapshot).freshContinuationIsolation !== "string") return false;
return (await readFreshContinuationReadiness(db, { id: "", companyId: input.companyId,
contextSnapshot: input.contextSnapshot ?? {} }, input.issueId)) !== null;
}
/**
* True when the fallback replaced this run and its continuation reuses none of
* the run's workspace: a new isolated workspace or an abandoned remote epoch.
* Holds that only protect the old executor's resources then no longer stand
* between the task and its next turn.
*/
export async function freshContinuationIsolatesRun(db: Db, input: { companyId: string; runId: string }): Promise<false | {
/** The isolated continuation is recorded but not yet delivered as a run. */
continuationPending: boolean;
}> {
const [record] = await db.select({ isolation: sql<string>`${issueRecoveryActions.evidence}#>>'{freshContinuation,isolation}'`,
delivery: sql<string | null>`${issueRecoveryActions.evidence}->>'continuationDelivery'` }).from(issueRecoveryActions).where(and(
eq(issueRecoveryActions.companyId, input.companyId),
sql`${issueRecoveryActions.evidence}#>>'{freshContinuation,sourceRunId}' = ${input.runId}`,
sql`${issueRecoveryActions.evidence}#>>'{freshContinuation,isolation}' in ('fresh_isolated_workspace', 'remote_epoch')`)).limit(1);
if (!record) return false;
// Only the recorded continuation carries the isolation marker that prepares
// the new workspace, so while it is pending it must be the task's next run.
return { continuationPending: record.isolation === "fresh_isolated_workspace" && record.delivery === "pending" };
}
export async function ensureFreshContinuation(db: Db, input: Input): Promise<FreshContinuationResult> {
return db.transaction(tx => continueFreshInTransaction(tx as unknown as Db, input));
}
@@ -474,9 +510,20 @@ export async function continueFreshInTransaction(tx: Db, input: Input): Promise<
// Saved user input (a message or a Retry the person pressed) continues the
// task through normal admission once nothing blocks it; a second, automatic
// continuation would run after it.
const [savedInput] = await tx.select({ id: agentWakeupRequests.id }).from(agentWakeupRequests).where(and(
const savedWakes = await tx.select({ id: agentWakeupRequests.id, payload: agentWakeupRequests.payload }).from(agentWakeupRequests).where(and(
eq(agentWakeupRequests.companyId, companyId), eq(agentWakeupRequests.status, "deferred_issue_execution"),
eq(agentWakeupRequests.requestedByActorType, "user"), sql`${agentWakeupRequests.payload}->>'issueId' = ${issue.id}`)).limit(1);
eq(agentWakeupRequests.requestedByActorType, "user"), sql`${agentWakeupRequests.payload}->>'issueId' = ${issue.id}`)).limit(5);
// Saved input continues the task only while it still holds something the
// agent has not received: a message the cut-off run already took into its
// turn is settled as delivered when promoted, and then nothing would run.
let savedInput: { id: string } | undefined;
for (const wake of savedWakes) {
const commentIds = queuedCommentIdsFromWakePayload(wake.payload);
if (!commentIds.length || (await undeliveredLegacyUserCommentIds(tx, companyId, issue.id, run.agentId, commentIds)).length) {
savedInput = wake;
break;
}
}
let delivery: "recorded" | "caller_owns" | "saved_input" = input.delivery === "caller_owns" ? "caller_owns"
: savedInput ? "saved_input" : "recorded";
@@ -492,10 +539,12 @@ export async function continueFreshInTransaction(tx: Db, input: Input): Promise<
// Never claim a stop that was not proven: record what isolates the successor.
const isolation: FreshContinuationIsolation = fenced || remoteSource ? "remote_epoch"
: writer === "ended" && !waitBounded ? "writer_ended" : "fresh_isolated_workspace";
// With no task workspace binding to move, only a run that carries the
// isolation in its context can avoid the shared directory: the recorded
// continuation runs first, and saved input follows it.
if (isolation === "fresh_isolated_workspace" && !issue.executionWorkspaceId && delivery === "saved_input") delivery = "recorded";
// Only a run that carries the isolation in its context prepares the new
// workspace (create_new). Saved input carries no such marker: on a task with
// no project it would be refused at admission once the binding is cleared,
// and otherwise could resolve the old directory. So the recorded
// continuation runs first, binds the new workspace, and saved input follows.
if (isolation === "fresh_isolated_workspace" && delivery === "saved_input") delivery = "recorded";
const unverified = await unverifiedActions(tx, run);
const note = freshContinuationNote({ runId: run.id, unverified,
workspace: isolation === "writer_ended" ? "kept" : "restored_last_save" });
+3 -1
View File
@@ -3,7 +3,7 @@ import { connectionIntentService } from "../connection-intents.js";
import { verifyBlockerStallWake } from "../recovery/blocker-stall-wakeups.js";
import { CONTEXT_COMPACTION_WAKE_REASON, ContextCompactionError, isContextCompactionRun, readContextCompactionRequest } from "../context-compaction.js";
import { isProcessRunAdmissionClosed } from "./run-admission.js";
import { continueFreshInTransaction, hasPendingFreshContinuation, settleCoalescedFreshContinuationHolds } from "../fresh-continuation-fallback.js";
import { continueFreshInTransaction, hasPendingFreshContinuation, isFreshIsolatedContinuationWake, settleCoalescedFreshContinuationHolds } from "../fresh-continuation-fallback.js";
import { refreshRunStartYieldIfStale } from "../run-start-handover.js";
import { deliveredQueuedCommentIds } from "../queued-comment-delivery.js";
import { hasAutomaticInteractionDeliveryHold, INTERACTION_DELIVERY_DENIED, recheckQueuedRunInteractions, withInteractionDeliveryOrigin } from "../agent-interaction-delivery.js";
@@ -4278,6 +4278,8 @@ export function createHeartbeatQueue(db: Db, dependencies: HeartbeatQueueDepende
reusableExecutionWorkspaceAvailable:
reuseRequest.existingExecutionWorkspaceAvailable,
hasResolvablePriorSessionWorkspace,
freshIsolatedContinuation: await isFreshIsolatedContinuationWake(tx as unknown as Db, {
companyId: issue.companyId, issueId: issue.id, contextSnapshot: enrichedContextSnapshot }),
})
) {
if (caseRequest) waitForCaseAdmission(WORKSPACE_WORKTREE_REQUIRES_PROJECT_CODE, WORKSPACE_WORKTREE_REQUIRES_PROJECT_MESSAGE);