Files
PaperClipAI/server/src/services/retry-workspace-continuity.test.ts
T
Nicky LeachandPaperclip b2edea5754 fix(workspaces): verify existing worktrees before fresh retries (#20876)
## Thinking Path

> - Paperclip manages work for AI agents.
> - Retry checks whether the task can keep its existing workspace.
> - A provider can complete checkout before Core receives the result.
> - Recreating a completed worktree costs time and can change its base
commit.
> - Core uses the same ownership and writer checks for local and
provider workspaces.
> - The provider verifies its checkout before Core creates a fresh
workspace.

## Linked Issues or Issue Description

**What happened?**

A retry can allocate a new provider worktree even though the original
checkout completed.

**Steps to reproduce**

Complete workspace checkout through a storage provider. Fail Runner
startup or lose the prepare reply. Retry the task.

**Expected behavior**

Reuse the same worktree when the provider confirms checkout and the
prior run can no longer write to it.

Related: #20865 addresses timeouts while the original prepare is in
progress. This PR checks an earlier checkout when the user retries.

## What Changed

- Check the prior task binding and writer state before asking a storage
provider to verify a checkout.
- Extend the existing prepare operation with an optional verification
mode. Providers opt in through their manifest.
- Return through normal workspace reuse after verification. Keep the
fresh retry fallback when verification cannot establish checkout.

## Verification

- Core workspace preparation, plugin workspace ownership, retry
continuity, and SDK workspace-provider suites passed locally (213
tests).
- Focused cases pass for provider verification, retry through heartbeat
dispatch, and real local Git checkout verification with no plugin
worker. The heartbeat test confirms that ordinary preparation and
adapter dispatch keep the original workspace, branch, and base commit.
- Core server typecheck passed before the final audit and timing edits.
GitHub CI covers the final revision and broader regressions.

## Risks

Provider verification adds one bounded call before a fresh retry. Older
providers keep their current behavior. Verification must preserve the
workspace, branch, and recorded base commit. The normal execution
preparation checks still run before admission. No database migration is
needed.

## Model Used

OpenAI GPT-6 (Codex), with source inspection and local test execution.

## 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 linked existing issues or described the issue in-PR
following the relevant issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links
- [x] My branch name describes the change and contains no internal issue
details
- [x] I have run targeted tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] All Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-11-11 18:04:18 +00:00

225 lines
17 KiB
TypeScript

import { execFile as execFileCallback } from "node:child_process";
import { randomUUID } from "node:crypto";
import { lstat, mkdtemp, readFile, rm, writeFile } from "node:fs/promises";
import { tmpdir } from "node:os";
import { join } from "node:path";
import { promisify } from "node:util";
import { and, eq } from "drizzle-orm";
import { afterAll, beforeAll, describe, expect, it } from "vitest";
import {
agentWakeupRequests, agents, applyPendingMigrations, companies, createDb, environmentLeases, executionWorkspaces,
heartbeatRuns, issueRecoveryActions, issues, nativeRunFinalizations, projects, workspaceOperations, type Db,
} from "@paperclipai/db";
import { getEmbeddedPostgresTestSupport, startEmbeddedPostgresTestDatabase } from "../__tests__/helpers/embedded-postgres.js";
import { resolveExternalTestDatabaseUrl } from "../__tests__/helpers/external-test-database.js";
import { admitExplicitNativeContinuation } from "./explicit-native-continuation.js";
import { readFreshWorkspaceReadiness, verifyRetryWorkspace } from "./fresh-workspace-continuation.js";
import { describeRetainedPriorWorkspace, parsePorcelainZ } from "./fresh-workspace-seed.js";
import { recordNativeLocalProcessStop } from "./native-local-process-stop.js";
import { createSystemRecovery, failSystemRecovery, readSystemRecovery, SystemRecoveryFailure } from "./system-recovery-state.js";
import { requestTaskRetry } from "./task-retry.js";
const execFile = promisify(execFileCallback);
const externalUrl = resolveExternalTestDatabaseUrl("PAPERCLIP_RETRY_WORKSPACE_TEST_DATABASE_URL");
const support = externalUrl ? { supported: true } : await getEmbeddedPostgresTestSupport();
/** R2: a Retry after recovery runs out must not drop the agent's work. */
describe.skipIf(!support.supported)("Retry workspace continuity after exhausted or blocked recovery", () => {
let database: Awaited<ReturnType<typeof startEmbeddedPostgresTestDatabase>>;
let db: Db;
let root: string;
beforeAll(async () => {
root = await mkdtemp(join(tmpdir(), "paperclip-retry-workspace-"));
if (externalUrl) {
await applyPendingMigrations(externalUrl);
database = { connectionString: externalUrl, cleanup: async () => {} } as typeof database;
} else database = await startEmbeddedPostgresTestDatabase("retry-workspace-continuity-");
db = createDb(database.connectionString);
}, 90_000);
afterAll(async () => {
await db?.$client.end();
await database?.cleanup();
if (root) await rm(root, { recursive: true, force: true });
});
async function git(cwd: string, ...args: string[]) {
return (await execFile("git", ["-C", cwd, ...args])).stdout.trim();
}
type Kind = "exhausted_generation" | "blocked_recovery" | "blocked_without_generation" | "duplicate_holds";
async function seed(kind: Kind) {
const companyId = randomUUID(), agentId = randomUUID(), issueId = randomUUID(), sourceRunId = randomUUID();
await db.insert(companies).values({ id: companyId, name: "Retry continuity", defaultResponsibleUserId: "board",
issuePrefix: `R${companyId.replaceAll("-", "").slice(0, 8)}` });
await db.insert(agents).values({ id: agentId, companyId, name: "Native", role: "engineer",
adapterType: "paperclip_runner", status: "idle", runtimeConfig: { heartbeat: { maxConcurrentRuns: 1 } } });
const [project] = await db.insert(projects).values({ companyId, name: "Repository" }).returning();
// A real primary checkout and the task's own git worktree with uncommitted edits.
const primary = join(root, `${issueId}-primary`);
await execFile("git", ["init", "--initial-branch=main", primary]);
await git(primary, "config", "user.email", "test@example.com");
await git(primary, "config", "user.name", "Retry test");
await writeFile(join(primary, "README.md"), "committed\n");
await git(primary, "add", ".");
await git(primary, "commit", "-m", "base");
const worktree = join(root, `${issueId}-worktree`);
const branchName = `TASK-${issueId.slice(0, 8)}`;
await git(primary, "worktree", "add", "-b", branchName, worktree, "main");
await writeFile(join(worktree, "README.md"), "committed\nedited before the fault\n");
await writeFile(join(worktree, "NOTES.md"), "untracked pre-fault work\n");
await db.insert(issues).values({ id: issueId, companyId, projectId: project.id, title: "Edit the repository",
status: "blocked", assigneeAgentId: agentId,
executionWorkspacePreference: "reuse_existing", executionWorkspaceSettings: { mode: "isolated_workspace" } });
const [workspace] = await db.insert(executionWorkspaces).values({ companyId, projectId: project.id, sourceIssueId: issueId,
mode: "isolated_workspace", strategyType: "git_worktree", name: branchName, status: "active", cwd: worktree,
providerType: "git_worktree", providerRef: worktree, baseRef: "main", branchName }).returning();
await db.update(issues).set({ executionWorkspaceId: workspace.id }).where(eq(issues.id, issueId));
const blockedKind = kind === "blocked_recovery" || kind === "blocked_without_generation";
const errorCode = blockedKind ? "native_restart_recovery_blocked" : "native_session_retry_exhausted";
const [source] = await db.insert(heartbeatRuns).values({ id: sourceRunId, companyId, agentId, nativeIssueId: issueId,
runtimeMode: "native", status: "failed", errorCode, processPid: 999_999_999,
contextSnapshot: { issueId }, startedAt: new Date(Date.now() - 120_000), finishedAt: new Date(Date.now() - 60_000) }).returning();
await db.insert(nativeRunFinalizations).values({ runId: sourceRunId, companyId, issueId,
phase: "terminal_failure", attempt: 3, failureCode: errorCode });
await db.insert(environmentLeases).values({ companyId, issueId, heartbeatRunId: sourceRunId, provider: "local",
status: "expired", leasePolicy: "ephemeral", executionWorkspaceId: workspace.id, releasedAt: new Date(Date.now() - 50_000),
cleanupStatus: "success" });
expect(await db.transaction(tx => recordNativeLocalProcessStop(tx as unknown as Db, source))).toBe(true);
// Recovery clears the process fields after it records the host's stop observation.
await db.update(heartbeatRuns).set({ processPid: null }).where(eq(heartbeatRuns.id, sourceRunId));
const generation = createSystemRecovery(sourceRunId);
const systemRecovery = kind === "exhausted_generation" || kind === "duplicate_holds"
? failSystemRecovery(generation, new SystemRecoveryFailure("source_unavailable", "use_alternative_strategy"))
: { ...generation, state: "complete" as const, phase: "settle_result_or_dispatch" as const };
const [action] = await db.insert(issueRecoveryActions).values({ companyId, sourceIssueId: issueId,
kind: "active_run_watchdog", ownerType: blockedKind ? "board" : "system", returnOwnerAgentId: agentId,
cause: blockedKind ? "native_restart_recovery_blocked" : "native_continuation_requires_reconciliation",
fingerprint: `native-restart:${sourceRunId}`, nextAction: "Retry the task.",
// A blocked restart without an admission hold records no generation.
evidence: kind === "blocked_without_generation" ? { runId: sourceRunId, reason: "ownership changed" }
: { runId: sourceRunId, systemRecovery } }).returning();
// Several holds can describe the same failed run; a historical one has no generation.
if (kind === "duplicate_holds") await db.insert(issueRecoveryActions).values({ companyId, sourceIssueId: issueId,
kind: "active_run_watchdog", ownerType: "system", returnOwnerAgentId: agentId, cause: "native_session_retry_exhausted",
status: "resolved", fingerprint: randomUUID(), nextAction: "Retry the task.", evidence: { runId: sourceRunId } });
return { companyId, agentId, issueId, source, workspace, worktree, primary, branchName, action };
}
async function retry(f: Awaited<ReturnType<typeof seed>>) {
const receipt = await requestTaskRetry({ db, companyId: f.companyId, issueId: f.issueId, actorId: "board",
observedRunId: f.source.id, idempotencyKey: randomUUID() });
const successorRunId = randomUUID();
const blocked: string[] = [];
const continuation = await db.transaction(async tx => {
await tx.select().from(issues).where(eq(issues.id, f.issueId)).for("update");
const admitted = await admitExplicitNativeContinuation({ db: tx as unknown as Db, companyId: f.companyId,
issueId: f.issueId, agentId: f.agentId, actorId: "board", actorType: "user", reason: "task_retry",
commentId: null, successorRunId, taskRetryRequestId: receipt.requestId!, onBlocked: reason => blocked.push(reason) });
await tx.insert(heartbeatRuns).values({ id: successorRunId, companyId: f.companyId, agentId: f.agentId,
nativeIssueId: f.issueId, runtimeMode: "native", status: "queued",
contextSnapshot: { issueId: f.issueId, taskRetryRequestId: receipt.requestId } });
await tx.update(agentWakeupRequests).set({ status: "coalesced", runId: successorRunId })
.where(eq(agentWakeupRequests.id, receipt.requestId!));
return admitted;
});
expect(blocked).toEqual([]);
expect(continuation).toMatchObject({ previousRunId: f.source.id, taskRetryRequestId: receipt.requestId });
const [successor] = await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, successorRunId));
return { successor, requestId: receipt.requestId!, readiness: await readFreshWorkspaceReadiness(db, successor) };
}
it.each(["exhausted_generation", "blocked_recovery", "blocked_without_generation", "duplicate_holds"] as const)(
"continues a %s Retry in the task's own git worktree with its uncommitted edits", async kind => {
const f = await seed(kind);
const { successor, requestId, readiness } = await retry(f);
expect(readiness).toEqual({ mode: "reuse_issue_workspace", retryIntentId: requestId,
sourceRunId: f.source.id, workspaceId: f.workspace.id, cwd: f.worktree });
// The retained directory is the same worktree: git history, branch and pre-fault files are intact.
expect((await lstat(join(readiness!.mode === "reuse_issue_workspace" ? (readiness!.cwd ?? "") : "", ".git"))).isFile()).toBe(true);
expect(await git(f.worktree, "rev-parse", "--abbrev-ref", "HEAD")).toBe(f.branchName);
expect(await readFile(join(f.worktree, "NOTES.md"), "utf8")).toBe("untracked pre-fault work\n");
expect(await readFile(join(f.worktree, "README.md"), "utf8")).toContain("edited before the fault");
// The task binding is unchanged and this attempt consumed the generation's alternative attempt once.
const [task] = await db.select().from(issues).where(eq(issues.id, f.issueId));
expect(task.executionWorkspaceId).toBe(f.workspace.id);
const [action] = await db.select().from(issueRecoveryActions).where(eq(issueRecoveryActions.id, f.action.id));
if (kind !== "blocked_without_generation") expect(readSystemRecovery(action.evidence)?.freshAttemptRunId).toBe(successor.id);
const [wake] = await db.select().from(agentWakeupRequests).where(eq(agentWakeupRequests.id, requestId));
expect((wake.payload?.taskRetry as Record<string, unknown>).strategy).toBe("continue");
expect(await readFreshWorkspaceReadiness(db, { ...successor, id: randomUUID() })).toBeNull();
}, 30_000);
it("verifies a local checkout after fresh fallback without a plugin worker", async () => {
const f = await seed("blocked_without_generation");
const { successor, requestId } = await retry(f);
expect(await verifyRetryWorkspace(db, undefined, successor, f.issueId, "board", {
mode: "fresh_isolated_workspace", retryIntentId: requestId, sourceRunId: f.source.id, retainedSourceIds: [],
})).toEqual({ mode: "reuse_issue_workspace", retryIntentId: requestId,
sourceRunId: f.source.id, workspaceId: f.workspace.id, cwd: f.worktree });
}, 30_000);
it("lists renamed paths from either porcelain status column once", () => {
expect(parsePorcelainZ("R new.md\0old.md\0 R moved.md\0before.md\0?? NOTES.md\0 M README.md\0")).toEqual([
"new.md", "moved.md", "NOTES.md", "README.md"]);
});
it("names the retained prior worktree and its uncommitted files for a fresh attempt", async () => {
const f = await seed("exhausted_generation");
const prior = await describeRetainedPriorWorkspace(db, { companyId: f.companyId, issueId: f.issueId,
sourceRunId: f.source.id, successorRunId: randomUUID(), freshCwd: join(root, "fresh") });
expect(prior).toEqual({ sourceRunId: f.source.id, path: f.worktree, branchName: f.branchName,
changedFilesKnown: true, changedFileCount: 2, changedFiles: expect.arrayContaining(["NOTES.md", "README.md"]) });
}, 30_000);
it("ignores a workspace operation orphaned by a restart of the stopped source", async () => {
const f = await seed("blocked_recovery");
// The server died during the source's preparation; no process can still run it.
await db.insert(workspaceOperations).values({ companyId: f.companyId, executionWorkspaceId: f.workspace.id,
heartbeatRunId: f.source.id, issueId: f.issueId, phase: "worktree_prepare", status: "running",
startedAt: new Date(Date.now() - 24 * 60 * 60_000) });
const { readiness } = await retry(f);
expect(readiness).toMatchObject({ mode: "reuse_issue_workspace", cwd: f.worktree });
}, 30_000);
it.each(["live_lease", "running_operation", "corrupt_git", "missing_directory", "remote_authority", "shared_workspace", "export_pending", "repair_required", "live_source_operation"] as const)(
"keeps the fresh isolated attempt when the prior workspace is unsafe: %s", async unsafe => {
const f = await seed("exhausted_generation");
if (unsafe === "live_lease") await db.insert(environmentLeases).values({ companyId: f.companyId, issueId: f.issueId,
provider: "local", status: "active", leasePolicy: "ephemeral", executionWorkspaceId: f.workspace.id });
if (unsafe === "running_operation") await db.insert(workspaceOperations).values({ companyId: f.companyId,
// A workspace-control job has no run and can predate the source settling.
executionWorkspaceId: f.workspace.id, issueId: f.issueId, phase: "worktree_prepare", status: "running",
startedAt: new Date(Date.now() - 24 * 60 * 60_000) });
if (unsafe === "corrupt_git") await writeFile(join(f.worktree, ".git"), "gitdir: /nonexistent/worktree\n");
if (unsafe === "missing_directory") await rm(f.worktree, { recursive: true, force: true });
if (unsafe === "remote_authority") await db.update(executionWorkspaces).set({ metadata: { remoteWorkspace: {
schema: "paperclip.remote-workspace.v1", epoch: 1, snapshotId: randomUUID(), source: { kind: "empty" },
location: null, writerRunId: null, dirty: false } } })
.where(eq(executionWorkspaces.id, f.workspace.id));
if (unsafe === "shared_workspace") await db.update(executionWorkspaces).set({ mode: "shared_workspace" })
.where(eq(executionWorkspaces.id, f.workspace.id));
if (unsafe === "repair_required") await db.update(heartbeatRuns).set({ resultJson: { workspaceRestoreFailure: "restore_failed" } })
.where(eq(heartbeatRuns.id, f.source.id));
// The source settled just now; an operation it started in this server process may still be running.
if (unsafe === "live_source_operation") {
await db.update(heartbeatRuns).set({ finishedAt: new Date() }).where(eq(heartbeatRuns.id, f.source.id));
await db.insert(workspaceOperations).values({ companyId: f.companyId, executionWorkspaceId: f.workspace.id,
heartbeatRunId: f.source.id, issueId: f.issueId, phase: "worktree_prepare", status: "running", startedAt: new Date(Date.now() - 1_000) });
}
if (unsafe === "export_pending") await db.update(issueRecoveryActions).set({ evidence: { ...f.action.evidence,
systemRecovery: { ...readSystemRecovery(f.action.evidence)!, state: "running" } } })
.where(eq(issueRecoveryActions.id, f.action.id));
const { readiness, successor } = await retry(f);
expect(readiness).toMatchObject({ mode: "fresh_isolated_workspace", sourceRunId: f.source.id });
// The fresh attempt names the retained directory only when reuse would also
// trust it as a source: task-owned, local, settled and with no other writer.
const prior = await describeRetainedPriorWorkspace(db, { companyId: f.companyId, issueId: f.issueId,
sourceRunId: f.source.id, successorRunId: successor.id, freshCwd: join(root, "fresh") });
if (unsafe === "corrupt_git") expect(prior).toEqual({ sourceRunId: f.source.id, path: f.worktree,
branchName: f.branchName, changedFilesKnown: false, changedFileCount: 0, changedFiles: [] });
else expect(prior).toBeNull();
const [task] = await db.select().from(issues).where(and(eq(issues.id, f.issueId), eq(issues.companyId, f.companyId)));
expect(task.executionWorkspaceId).toBe(f.workspace.id);
}, 30_000);
});