From 6cb0dd0c6bc911f4cf98d1e39c5a78c0de29bdfd Mon Sep 17 00:00:00 2001 From: Dotta <34892728+cryppadotta@users.noreply.github.com> Date: Thu, 12 Nov 2026 21:10:47 -0600 Subject: [PATCH] fix(recovery): continue twice-interrupted tasks and waiting follow-ups after a release swap (#21250) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit ## 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 --- .../execution-workspace-policy.test.ts | 8 ++ .../wake-queue/adapters/postgres.test.ts | 82 ++++++++++++++++++- .../modules/wake-queue/adapters/postgres.ts | 31 ++++++- .../services/execution-workspace-policy.ts | 4 + .../fresh-continuation-fallback.test.ts | 26 +++++- .../services/fresh-continuation-fallback.ts | 61 ++++++++++++-- server/src/services/heartbeat/queue.ts | 4 +- 7 files changed, 200 insertions(+), 16 deletions(-) diff --git a/server/src/__tests__/execution-workspace-policy.test.ts b/server/src/__tests__/execution-workspace-policy.test.ts index e44ea327f3a..6253cba8841 100644 --- a/server/src/__tests__/execution-workspace-policy.test.ts +++ b/server/src/__tests__/execution-workspace-policy.test.ts @@ -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({ diff --git a/server/src/modules/wake-queue/adapters/postgres.test.ts b/server/src/modules/wake-queue/adapters/postgres.test.ts index 01a6c659415..aac74f1f3bc 100644 --- a/server/src/modules/wake-queue/adapters/postgres.test.ts +++ b/server/src/modules/wake-queue/adapters/postgres.test.ts @@ -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", diff --git a/server/src/modules/wake-queue/adapters/postgres.ts b/server/src/modules/wake-queue/adapters/postgres.ts index 95341f6bf18..ff582a93cc6 100644 --- a/server/src/modules/wake-queue/adapters/postgres.ts +++ b/server/src/modules/wake-queue/adapters/postgres.ts @@ -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 && diff --git a/server/src/services/execution-workspace-policy.ts b/server/src/services/execution-workspace-policy.ts index 4a7f9c9fa33..1b27f8fcc04 100644 --- a/server/src/services/execution-workspace-policy.ts +++ b/server/src/services/execution-workspace-policy.ts @@ -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; diff --git a/server/src/services/fresh-continuation-fallback.test.ts b/server/src/services/fresh-continuation-fallback.test.ts index 1cc5121b76c..8aeb661c102 100644 --- a/server/src/services/fresh-continuation-fallback.test.ts +++ b/server/src/services/fresh-continuation-fallback.test.ts @@ -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)); diff --git a/server/src/services/fresh-continuation-fallback.ts b/server/src/services/fresh-continuation-fallback.ts index e5e5ab91c6c..adc6eb81df5 100644 --- a/server/src/services/fresh-continuation-fallback.ts +++ b/server/src/services/fresh-continuation-fallback.ts @@ -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 | 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 { + const [record] = await db.select({ isolation: sql`${issueRecoveryActions.evidence}#>>'{freshContinuation,isolation}'`, + delivery: sql`${issueRecoveryActions.evidence}->>'continuationDelivery'` }).from(issueRecoveryActions).where(and( + eq(issueRecoveryActions.companyId, input.companyId), + sql`${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 { 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" }); diff --git a/server/src/services/heartbeat/queue.ts b/server/src/services/heartbeat/queue.ts index f4d53e04744..fa0ff7b8b59 100644 --- a/server/src/services/heartbeat/queue.ts +++ b/server/src/services/heartbeat/queue.ts @@ -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);