fix(workspaces): preserve import verification across process restarts (#20311)

## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - Online import preserves old local files before the web server can
use a disposable filesystem.
> - A parent import must authenticate saved child archives before it can
omit their directories.
> - The operation journal preserves import identity, but completed child
checks existed only in process memory.
> - A container restart therefore repeated downloads, decryption and
restoration that had already succeeded.
> - This pull request saves a versioned verification receipt on each
immutable child snapshot.
> - A replacement process can reuse those checks while it reacquires
locks and verifies current local files.

## Linked Issues or Issue Description

Refs #19854 for verification reuse within a live source fence. Refs
#19873 for cohort import. Refs #20298 for the separate remote-hold
admission fix. Searched related workspace verification and restart PRs;
none provides durable child verification.

**What happened?**

Replacing a server container interrupts a long online import. The
original operation and snapshot survive, but a later invocation
downloads, decrypts and restores completed child archives again. The
current cache ends with the source-lock owner.

**Expected behavior**

Completed immutable checks survive process and scratch loss. Resume
keeps the same operation and snapshot, reuses only correctly bound
verification, and publishes once. Current local files and Git state must
still be checked under fresh source locks.

**Steps to reproduce**

1. Import two nested child workspaces, then start the parent import.
2. Kill the importer after a child verification finishes. Remove its
scratch directory.
3. Resume in a new process. Before this fix, the completed child is
downloaded and restored again.

**Paperclip version or commit**

The regression fails on the previous migration verifier. The final
branch includes master at `189fefb99183953be37f0d1084275fbb0f35d6e2`.

**Deployment mode**

Managed Linux server with PostgreSQL and explicit S3 storage
configuration.

## What Changed

- Add optional opaque storage namespace identity. S3 binds it to the
explicit endpoint, region, bucket, normalized prefix and addressing
style. Implicit endpoints retain full verification.
- Save a versioned child-snapshot receipt only after canonical crypto,
manifest and file restoration succeed and a second HEAD confirms the
same strong ETag and size.
- Bind reuse to the complete snapshot reference, key material, child
lineage, source operation and storage namespace. Require a fresh strong
HEAD and canonical validation that the current secret provider can
unwrap the saved snapshot key.
- Keep current directory/Git capture and source-owner checks on every
invocation. Read-only verifier callers do not write checkpoints.
- Merge checkpoint metadata under a short row lock without changing
parent journal hashes, snapshot retention or update timestamps. Bound
reads and writes. Lock timeouts, statement timeouts, deadlocks and
serialization conflicts are safe cache misses. Other errors remain
fatal.
- Extend real PostgreSQL/Git/archive tests with process death, scratch
removal, resume, changed bindings and single publication.

## Verification

- Baseline RED: two import invocations read each child twice. With
durable receipts, each child is read once across both invocations.
- Real process regression passed on the final rebased head: SIGKILL the
owned importer after one checkpoint, delete all scratch, and resume in a
new Node process. It retains the same operation and snapshot, reads each
child once across both processes, and records one publication.
- 53 focused cases passed in 104.86 seconds on the final production
bytes before rebase. They cover existing nested/cohort behavior,
namespace identity, changed source/reference/key/lineage/ETag, unknown
proof versions, weak validators and bounded row contention.
- After clean rebase and stronger scratch/snapshot assertions, the real
process case passed again in 18.54 seconds. Server TypeScript, module
boundaries and diff checks passed on final head
`9da68ee7ce4a778201ef85e21a11c49eeea7bc16`. Required CI supplies
final-head broad coverage.
- The initial restricted-network dependency setup failed; unchanged
preflight passed with normal network access. Initial TypeScript failures
in test pipe assertions were corrected. No failed attempt is counted as
a pass.
- Review found a missing current-master-key check. A real restart
regression with all child checkpoints complete reproduced unsafe
publication after switching to a different valid master key. Follow-up
`7ea92c2abd9ffdb2e7dd81cecf62c8fc7a6e74a5` delegates to the canonical
snapshot-key owner before both durable and in-memory cache hits.
- The fixed process regressions passed 2/2. Ten key-authority and
existing identity/source guard cases passed. The wrong-key process now
refuses with zero additional child payload GETs, no publication, and
unchanged original journal/snapshot rows. Server typecheck, module
boundaries and diff checks passed on the follow-up.
- Review also found that transient database failures could make the
optional checkpoint cache stop a valid import. Read and save now fall
back only for SQLSTATE `55P03`, `57014`, `40P01`, and `40001`.
Connection loss, identity changes, crypto failures, programming errors
and cancellation remain fatal. Reads use a read-only transaction with
one-second lock and eight-second statement limits.
- The checkpoint database-fault regression failed two cases before the
fix. All four recoverable import cases then passed. The final selected
run passed 21 cases, including fatal-error and cancellation controls,
real row contention, both process-restart cases and key/identity guards.
Server typecheck and boundaries passed. Fault injection is limited to
the checkpoint database boundary; the importer, PostgreSQL, Git,
archives, storage and publication remain real.
- Latest master was merged without rewriting published history. Six real
import/process-restart cases passed again on final head
`8c8e4213a680ebb5c8d6864416bc3a4f271e1e76` in 38.03 seconds. [Required
CI](https://github.com/paperclipai/paperclip/actions/runs/42658248222)
finished with 54 successful checks and four intentional skips; Snyk
passed. [Fresh
review](https://github.com/paperclipai/paperclip/pull/20311#issuecomment-6745354381)
is 5/5 with all four threads addressed and resolved. Hosted verification
after normal release delivery remains pending.

## Risks

- A stale receipt could skip authentication of a different object.
Complete reference/lineage/namespace binding, fresh strong HEAD and
row-locked save validation prevent that reuse. Missing or unsupported
hints trigger full verification.
- Mutable local files can change between attempts. Their source locks
and fresh captures remain required; a checkpoint never certifies current
local bytes.
- Checkpoint persistence can contend with other snapshot owners. It has
one-second lock and eight-second statement limits; the four classified
transient SQL errors retain the normal fully verified path. Other
database errors stop the import so lost source ownership cannot be
hidden.
- No schema, archive format, public API or saved-run identity change.
Older snapshots without receipts remain readable and get full
verification. Existing data and recovery obligations stay intact.

## Model Used

OpenAI Codex, GPT-6.1 Sol (`gpt-6.1-sol`), with reasoning, repository
tools and code execution. Context-window size is not exposed by the
runtime.

## 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
- [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
- [x] I will address all Greptile and reviewer comments before
requesting merge
This commit is contained in:
Dotta authored and GitHub committed 2026-11-09 11:03:48 -06:00
1 parent 0cf70c9d0c
commit 1b7fe33432
10 files changed
+416 -7

No files matched your search

@@ -100,3 +100,20 @@ serialized remote function, compare archive bytes and digests, remove every
original source, restore uncommitted/staged work, and run fsck. No hosted import,
hold, cancellation, provider call, or shared-cache Git command is part of these
local tests.
The online migration owner can checkpoint completed immutable nested-snapshot
verification in the child snapshot ledger. This optional versioned proof binds
the complete encrypted reference, child lineage, source operation and a known
S3 endpoint/region/bucket/prefix/style namespace. A new process can reuse it only
after fresh strong-ETag/length, current ledger/source-fence checks and canonical
metadata-only unwrapping of each child key under the current master key. This
key check also runs for in-memory hits; it returns no secret or payload. The proof
is published only after canonical decryption, manifest/file restore and a stable
post-verification HEAD. Missing or unknown proof versions, implicit provider
endpoints and weak or changed validators require full verification. Existing
records need no migration. No plaintext scratch, parent journal CAS, authority,
retention deadline or local/Git source proof is reused. Every resumed invocation
reacquires the source locks and captures current local bytes again. Read-only
verifier callers do not opt into checkpoint publication.
A change that strengthens immutable archive verification must advance the proof
version so earlier verification cannot bypass the new checks.
@@ -0,0 +1,24 @@
import { describe, expect, it } from "vitest";
import { createS3StorageProvider } from "../storage/s3-provider.js";
import { createStorageService } from "../storage/service.js";
const config = { bucket: "fixture", region: "us-east-1", endpoint: "https://storage.example.test", prefix: "root", forcePathStyle: true };
describe("immutable object namespace", () => {
it("propagates the opaque resource identity without credential material", () => {
const provider = createS3StorageProvider(config), service = createStorageService(provider);
expect(provider.immutableObjectNamespace).toMatch(/^[a-f0-9]{64}$/);
expect(service.immutableObjectNamespace).toBe(provider.immutableObjectNamespace);
expect(createS3StorageProvider({ ...config, endpoint: "https://storage.example.test/", prefix: "/root/" }).immutableObjectNamespace)
.toBe(provider.immutableObjectNamespace);
});
it.each([
{ bucket: "other" }, { region: "us-west-2" }, { endpoint: "https://other.example.test" },
{ prefix: "other" }, { forcePathStyle: false },
])("separates a changed resource location %j", change => {
expect(createS3StorageProvider({ ...config, ...change }).immutableObjectNamespace)
.not.toBe(createS3StorageProvider(config).immutableObjectNamespace);
});
it("does not attest an endpoint selected implicitly by SDK profiles or environment", () => {
expect(createS3StorageProvider({ bucket: "fixture", region: "us-east-1" }).immutableObjectNamespace).toBeUndefined();
});
});
@@ -12,7 +12,8 @@ import { localWorkspaceSourcePaths } from "./execution-workspace-source-roots.js
import { offlineMigrationExecutionWhere } from "./offline-migration-execution.js";
import type { StorageService } from "../storage/types.js";
import { captureRemoteWorkspaceDirectory, restoreRemoteWorkspaceArchive } from "./remote-workspace-snapshot-archive.js";
import { getWorkspaceSnapshot, loadWorkspaceSnapshotArchive, type WorkspaceSnapshotRef } from "./remote-workspace-snapshot-storage.js";
import { assertWorkspaceSnapshotKeyReadable, getWorkspaceSnapshot, loadWorkspaceSnapshotArchive, type WorkspaceSnapshotRef } from "./remote-workspace-snapshot-storage.js";
import { nestedWorkspaceVerificationCheckpoint } from "./remote-workspace-verification-checkpoint.js";
type ReadDb = Pick<Db, "select">;
type Workspace = typeof executionWorkspaces.$inferSelect;
@@ -304,8 +305,9 @@ export async function inspectImportedWorkspaceSourcePaths(db: ReadDb, workspace:
* the parent archive. The snapshot owner retains the resulting dependencies. */
export function createNestedWorkspaceImportVerifier(input: {
db: Db; storage: StorageService; companyId: string; sourceLock: SourceTransaction;
projectIds: readonly string[]; issueIds: readonly string[]; signal?: AbortSignal;
projectIds: readonly string[]; issueIds: readonly string[]; signal?: AbortSignal; persistVerification?: boolean;
}) {
const checkpoint = input.persistVerification ? nestedWorkspaceVerificationCheckpoint(input.db, input.storage.immutableObjectNamespace, input.signal) : undefined;
const stop = new AbortController(), signal = input.signal ? AbortSignal.any([input.signal, stop.signal]) : stop.signal;
const proofs = new Map<string, { etag: string; checked: Promise<boolean> }>();
let closed = false, current: Promise<ImportedWorkspaceChild[]> | undefined;
@@ -329,15 +331,28 @@ export function createNestedWorkspaceImportVerifier(input: {
&& head.etag.replace(/^"|"$/g, "").trim() && !/^W\//i.test(head.etag) ? head.etag : null;
const key = createHash("sha256").update(JSON.stringify([reference, children, operationId])).digest("hex");
const previous = proofs.get(key);
if (previous && etag === previous.etag
&& await previous.checked) return;
if (previous && etag === previous.etag && await previous.checked) {
await assertWorkspaceSnapshotKeyReadable(reference);
return;
}
proofs.delete(key);
if (etag && await checkpoint?.read({ reference, children, operationId, etag })) {
await assertWorkspaceSnapshotKeyReadable(reference);
await assertOwner();
proofs.set(key, { etag, checked: Promise.resolve(true) });
return;
}
const checked = (async () => {
await loadWorkspaceSnapshotArchive({ storage: input.storage, reference, targetPath: archivePath, signal });
await restoreRemoteWorkspaceArchive({ archivePath, targetDir: restored, expectedManifestDigest: reference.manifestDigest, signal });
if (!etag) return false;
const after = await objectHead(reference);
return after !== null && after.exists && after.etag === etag && after.contentLength === reference.byteSize;
const stable = after !== null && after.exists && after.etag === etag && after.contentLength === reference.byteSize;
if (stable && checkpoint) {
await assertOwner();
await checkpoint.save({ reference, children, operationId, etag });
}
return stable;
})();
const proof = etag ? { etag, checked } : null;
if (proof) proofs.set(key, proof);
@@ -418,7 +418,7 @@ async function lockMigrationRecovery(lock: MigrationTransaction, plans: Migratio
function createMigrationChildVerifier(input: MigrationInput, onlineSource: OnlineWorkspaceSource | undefined, sourceLock: MigrationTransaction) {
if (onlineSource?.kind !== "git") return undefined;
const verifier = createNestedWorkspaceImportVerifier({ db: input.db, storage: input.storage, companyId: input.companyId,
sourceLock, projectIds: onlineSource.projectIds, issueIds: onlineSource.issueIds, signal: input.signal });
sourceLock, projectIds: onlineSource.projectIds, issueIds: onlineSource.issueIds, signal: input.signal, persistVerification: true });
let pending = Promise.resolve();
return {
prepare(options: Parameters<typeof prepareNestedWorkspaceImports>[0]) {
@@ -1,4 +1,5 @@
import { execFile } from "node:child_process";
import { execFile, spawn } from "node:child_process";
import { fileURLToPath } from "node:url";
import { createHash, randomUUID } from "node:crypto";
import fs from "node:fs/promises";
import os from "node:os";
@@ -16,6 +17,9 @@ import { createS3StorageProvider } from "../storage/s3-provider.js";
import { createStorageService } from "../storage/service.js";
import type { StorageService } from "../storage/types.js";
import { migrateRemoteWorkspace } from "./remote-workspace-migration.js";
import { getWorkspaceSnapshot } from "./remote-workspace-snapshot-storage.js";
import { nestedWorkspaceVerificationCheckpoint } from "./remote-workspace-verification-checkpoint.js";
import * as checkpointOwner from "./remote-workspace-verification-checkpoint.js";
import * as workspaceArchive from "./remote-workspace-snapshot-archive.js";
import { createNestedWorkspaceImportVerifier } from "./remote-workspace-migration-tree.js";
import { inspectOnlineWorkspaceSource } from "./online-workspace-source.js";
@@ -33,6 +37,7 @@ describe.skipIf(!support.supported)("bounded nested workspace verification", ()
let database: Awaited<ReturnType<typeof startEmbeddedPostgresTestDatabase>>, db: ReturnType<typeof createDb>, home: string, storage: StorageService, server: Server;
let headEtags: Map<string, string | null>;
let headFailures: Map<string, () => boolean>;
let storageNamespace: string | undefined;
beforeAll(async () => { database = await startEmbeddedPostgresTestDatabase("nested-verification-"); db = createDb(database.connectionString, { maxConnections: 8 }); }, 90_000);
afterAll(async () => { await db?.$client.end({ timeout: 1 }); await database?.cleanup(); });
beforeEach(async () => {
@@ -61,6 +66,7 @@ describe.skipIf(!support.supported)("bounded nested workspace verification", ()
const heads = createStorageService(createS3StorageProvider({ bucket: "fixture", region: "us-east-1",
endpoint: `http://127.0.0.1:${address.port}`, forcePathStyle: true }));
vi.spyOn(storage, "headObject").mockImplementation(heads.headObject.bind(heads));
storageNamespace = heads.immutableObjectNamespace;
});
afterEach(async () => {
vi.restoreAllMocks(); await db.delete(activityLog); await db.delete(workspaceOperations); await db.delete(executionWorkspaceSnapshots);
@@ -113,6 +119,257 @@ describe.skipIf(!support.supported)("bounded nested workspace verification", ()
}
const parent = (f: Awaited<ReturnType<typeof imported>>, signal = AbortSignal.timeout(50_000)) =>
migrateRemoteWorkspace({ db, storage, companyId: f.companyId, workspaceId: f.parent.id, apply: true, signal });
function checkpointFailure(phase: "read" | "save", failure: Error, beforeFailure?: () => void) {
const original = checkpointOwner.nestedWorkspaceVerificationCheckpoint;
let faults = 0;
const decorate = (target: Db): Db => new Proxy(target, { get(object, key) {
if (key === (phase === "read" ? "select" : "update")) return () => { faults++; beforeFailure?.(); throw failure; };
if (key === "transaction") return (callback: (tx: Db) => Promise<unknown>, options: unknown) =>
object.transaction(tx => callback(decorate(tx as unknown as Db)), options as never);
const value = Reflect.get(object, key, object); return typeof value === "function" ? value.bind(object) : value;
} });
vi.spyOn(checkpointOwner, "nestedWorkspaceVerificationCheckpoint").mockImplementation((source, namespace, signal) =>
original(decorate(source), namespace, signal));
return () => faults;
}
it.each([{ phase: "read", code: "57014" }, { phase: "save", code: "57014" },
{ phase: "save", code: "40P01" }, { phase: "read", code: "40001" }])(
"keeps import resumable after checkpoint $phase $code failure", async ({ phase, code }) => {
const f = await imported(), get = storage.getObject.bind(storage), counts = new Map<string, number>();
storage = { ...storage, immutableObjectNamespace: storageNamespace };
vi.spyOn(storage, "getObject").mockImplementation(async (...args) => {
if (f.keys.has(args[1])) counts.set(args[1], (counts.get(args[1]) ?? 0) + 1);
return get(...args);
});
if (phase === "read") {
const capture = workspaceArchive.captureRemoteWorkspaceDirectory;
const stop = vi.spyOn(workspaceArchive, "captureRemoteWorkspaceDirectory").mockImplementation(async input => {
if (input.cwd === f.root) throw new Error("simulated_container_restart");
return capture(input);
});
await expect(parent(f)).rejects.toThrow("simulated_container_restart"); stop.mockRestore();
expect((await db.select().from(executionWorkspaceSnapshots)).filter(row => row.metadata?.nestedImportVerification)).toHaveLength(f.keys.size);
counts.clear();
}
const pgError = Object.assign(new Error("injected_checkpoint_database_failure"), { code });
const failure = code === "40001" ? new Error("checkpoint_query_failed", { cause: pgError }) : pgError;
const faults = checkpointFailure(phase as "read" | "save", failure);
await expect(parent(f)).resolves.toMatchObject({ applied: true });
expect(faults()).toBeGreaterThan(0); expect([...counts.values()]).toEqual([1, 1]);
if (phase === "save") expect((await db.select().from(executionWorkspaceSnapshots)).filter(row => row.metadata?.nestedImportVerification)).toEqual([]);
const operations = await db.select().from(workspaceOperations).where(eq(workspaceOperations.executionWorkspaceId, f.parent.id));
expect(operations).toHaveLength(1); expect(operations[0].status).toBe("succeeded");
expect((await db.select().from(activityLog)).filter(row => row.action === "workspace.remote_storage_migrated" && row.entityId === f.parent.id)).toHaveLength(1);
for (const child of f.leaves) expect(await fs.readFile(path.join(child.cwd!, "untracked.txt"), "utf8")).toBe(`${child.name} untracked work\n`);
}, 60_000);
it.each([{ phase: "read", code: "42P01", message: "checkpoint_table_missing" },
{ phase: "save", code: "08006", message: "checkpoint_connection_lost" },
{ phase: "save", code: null, message: "nested_workspace_snapshot_changed" },
{ phase: "read", code: null, message: "unexpected_checkpoint_state" }])(
"refuses checkpoint $phase failure $code without hiding the cause", async ({ phase, code, message }) => {
const f = await imported(); storage = { ...storage, immutableObjectNamespace: storageNamespace };
const failure = Object.assign(message === "unexpected_checkpoint_state" ? new TypeError(message) : new Error(message), code ? { code } : {});
const faults = checkpointFailure(phase as "read" | "save", failure);
await expect(parent(f)).rejects.toBe(failure); expect(faults()).toBeGreaterThan(0);
expect(readRemoteWorkspaceAuthority((await db.select().from(executionWorkspaces).where(eq(executionWorkspaces.id, f.parent.id)))[0].metadata)).toBeNull();
expect((await db.select().from(activityLog)).filter(row => row.action === "workspace.remote_storage_migrated" && row.entityId === f.parent.id)).toHaveLength(0);
for (const child of f.leaves) expect(await fs.readFile(path.join(child.cwd!, "untracked.txt"), "utf8")).toBe(`${child.name} untracked work\n`);
}, 60_000);
it("preserves cancellation over a retryable checkpoint database error", async () => {
const f = await imported(), controller = new AbortController(), stopped = new Error("cancelled_during_checkpoint_read");
storage = { ...storage, immutableObjectNamespace: storageNamespace };
const reference = (await getWorkspaceSnapshot(db, { companyId: f.companyId, workspaceId: f.snapshots[0].executionWorkspaceId, snapshotId: f.snapshots[0].id }))!;
const original = checkpointOwner.nestedWorkspaceVerificationCheckpoint;
checkpointFailure("read", Object.assign(new Error("query_cancelled"), { code: "57014" }), () => controller.abort(stopped));
const checkpoint = checkpointOwner.nestedWorkspaceVerificationCheckpoint(db, storageNamespace, controller.signal)!;
await expect(checkpoint.read({ reference, children: [], operationId: f.snapshots[0].sourceOperationId, etag: '"fixture"' })).rejects.toBe(stopped);
expect(original(db, storageNamespace, controller.signal)).toBeDefined();
}, 60_000);
it("treats checkpoint row contention as a bounded optional cache miss", async () => {
const f = await imported(), snapshot = f.snapshots[0];
const reference = (await getWorkspaceSnapshot(db, { companyId: f.companyId, workspaceId: snapshot.executionWorkspaceId, snapshotId: snapshot.id }))!;
const head = await storage.headObject(f.companyId, reference.objectKey), checkpoint = nestedWorkspaceVerificationCheckpoint(db, storageNamespace)!;
const binding = { reference, children: [], operationId: snapshot.sourceOperationId, etag: head.etag! };
await db.transaction(async tx => {
await tx.select().from(executionWorkspaceSnapshots).where(eq(executionWorkspaceSnapshots.id, snapshot.id)).for("update");
const started = performance.now();
await expect(checkpoint.save(binding)).resolves.toBeUndefined();
expect(performance.now() - started).toBeLessThan(2500);
expect(await checkpoint.read(binding)).toBe(false);
});
expect((await db.select().from(executionWorkspaceSnapshots).where(eq(executionWorkspaceSnapshots.id, snapshot.id)))[0]).toEqual(snapshot);
}, 10_000);
it.skipIf(process.platform === "win32").each(["partial", "changed-master-key"] as const)("resumes a killed import with %s checkpoints and the current key authority", async mode => {
const f = await imported(), repoRoot = fileURLToPath(new URL("../../../", import.meta.url));
const address = server.address(); if (!address || typeof address === "string") throw new Error("fixture_address_missing");
const keys = [...f.keys], endpoint = `http://127.0.0.1:${address.port}`, temporary = path.join(home, "process-scratch");
await fs.mkdir(temporary);
const body = String.raw`
const {createDb,closeRegisteredClients}=await import('@paperclipai/db');
const {createLocalDiskStorageProvider}=await import('./src/storage/local-disk-provider.ts');
const {createS3StorageProvider}=await import('./src/storage/s3-provider.ts');
const {createStorageService}=await import('./src/storage/service.ts');
const {migrateRemoteWorkspace}=await import('./src/services/remote-workspace-migration.ts');
const input=JSON.parse(process.env.IMPORT_FIXTURE);
const db=createDb(process.env.DATABASE_URL,{maxConnections:8});
const local=createStorageService(createLocalDiskStorageProvider(input.objects));
const heads=createStorageService(createS3StorageProvider({bucket:'fixture',region:'us-east-1',endpoint:input.endpoint,forcePathStyle:true}));
const storage={...local,immutableObjectNamespace:heads.immutableObjectNamespace,headObject:heads.headObject.bind(heads),
async getObject(companyId,key,options){
if(input.pauseKey===key)await new Promise(()=>{});
if(input.keys.includes(key))process.send?.({read:key});
return local.getObject(companyId,key,options);
},async putFile(value){
if(input.pauseParentPut)await new Promise(()=>{});
return local.putFile(value);
}};
try{const result=await migrateRemoteWorkspace({db,storage,companyId:input.companyId,workspaceId:input.workspaceId,
apply:true,signal:AbortSignal.timeout(30000)});process.send?.({result});}
catch(error){process.send?.({failed:true,message:error.message});process.exitCode=1;}
finally{await closeRegisteredClients();process.exit(process.exitCode??0);}
`;
const launch = (pauseKey: string | null, pauseParentPut = false, masterKey = process.env.PAPERCLIP_SECRETS_MASTER_KEY) => {
const reads: string[] = [], results: unknown[] = [], failures: string[] = [];
const child = spawn(process.execPath, ["--import", path.join(repoRoot, "cli/node_modules/tsx/dist/loader.mjs"),
"--input-type=module", "--eval", body], { cwd: path.join(repoRoot, "server"), detached: true,
env: { ...process.env, PAPERCLIP_SECRETS_MASTER_KEY: masterKey, TMPDIR: temporary, DATABASE_URL: database.connectionString, IMPORT_FIXTURE: JSON.stringify({
companyId: f.companyId, workspaceId: f.parent.id, objects: path.join(home, "objects"), endpoint, keys, pauseKey, pauseParentPut }) },
stdio: ["ignore", "pipe", "pipe", "ipc"] });
let stderr = ""; child.stderr!.on("data", bytes => { stderr += bytes; });
child.stdout!.resume(); child.on("message", (message: { read?: string; result?: unknown; failed?: boolean; message?: string }) => {
if (message.read) reads.push(message.read); if (message.result) results.push(message.result);
if (message.failed && message.message) failures.push(message.message);
});
const closed = new Promise<{ code: number | null; signal: string | null }>((resolve, reject) => {
child.once("error", reject); child.once("close", (code, signal) => resolve({ code, signal }));
});
return { child, reads, results, failures, closed, error: () => stderr };
};
const first = mode === "partial" ? launch(keys[1]) : launch(null, true);
try {
await Promise.race([vi.waitFor(async () => {
const [snapshot] = await db.select().from(executionWorkspaceSnapshots).where(eq(executionWorkspaceSnapshots.id, f.snapshots[0].id));
const completed = (await db.select().from(executionWorkspaceSnapshots)).filter(row => row.metadata?.nestedImportVerification);
expect(completed).toHaveLength(mode === "partial" ? 1 : keys.length); expect(snapshot).toBeDefined();
}, { timeout: 15_000, interval: 25 }), first.closed.then(result => { throw new Error(`import_ended_before_checkpoint:${result.code}:${first.error()}`); })]);
process.kill(-first.child.pid!, "SIGKILL");
expect(await first.closed).toMatchObject({ signal: "SIGKILL" });
} finally {
if (first.child.exitCode === null && first.child.signalCode === null) { process.kill(-first.child.pid!, "SIGKILL"); await first.closed; }
}
const [before] = await db.select().from(workspaceOperations).where(eq(workspaceOperations.executionWorkspaceId, f.parent.id));
expect(before.status).toBe("running");
const [beforeSnapshot] = await db.select().from(executionWorkspaceSnapshots).where(eq(executionWorkspaceSnapshots.executionWorkspaceId, f.parent.id));
await fs.rm(temporary, { recursive: true, force: true }); await fs.mkdir(temporary);
const snapshotsBefore = await db.select().from(executionWorkspaceSnapshots);
const second = launch(null, false, mode === "changed-master-key" ? "62".repeat(32) : undefined), result = await second.closed;
if (mode === "changed-master-key") {
expect(result, second.error()).toMatchObject({ code: 1, signal: null });
expect(second.failures).toEqual([expect.stringContaining("Secret decryption failed")]);
expect(second.reads).toEqual([]); expect(second.results).toEqual([]);
expect((await db.select().from(workspaceOperations).where(eq(workspaceOperations.executionWorkspaceId, f.parent.id)))[0]).toEqual(before);
expect(await db.select().from(executionWorkspaceSnapshots)).toEqual(snapshotsBefore);
expect(readRemoteWorkspaceAuthority((await db.select().from(executionWorkspaces).where(eq(executionWorkspaces.id, f.parent.id)))[0].metadata)).toBeNull();
expect((await db.select().from(activityLog)).filter(row => row.action === "workspace.remote_storage_migrated" && row.entityId === f.parent.id)).toHaveLength(0);
return;
}
expect(result, second.error()).toMatchObject({ code: 0, signal: null });
expect(second.results).toEqual([expect.objectContaining({ applied: true, snapshotId: beforeSnapshot.id })]);
expect([...first.reads, ...second.reads].sort()).toEqual([...keys].sort());
const operations = await db.select().from(workspaceOperations).where(eq(workspaceOperations.executionWorkspaceId, f.parent.id));
expect(operations).toHaveLength(1); expect(operations[0]).toMatchObject({ id: before.id, status: "succeeded" });
expect((await db.select().from(activityLog)).filter(row => row.action === "workspace.remote_storage_migrated" && row.entityId === f.parent.id)).toHaveLength(1);
}, 60_000);
it("rechecks the current master key before reusing an in-memory child proof", async () => {
const f = await imported(), get = storage.getObject.bind(storage), counts = new Map<string, number>();
vi.spyOn(storage, "getObject").mockImplementation(async (...args) => {
if (f.keys.has(args[1])) counts.set(args[1], (counts.get(args[1]) ?? 0) + 1);
return get(...args);
});
const capture = workspaceArchive.captureRemoteWorkspaceDirectory; let parentCaptures = 0;
vi.spyOn(workspaceArchive, "captureRemoteWorkspaceDirectory").mockImplementation(async input => {
const result = await capture(input);
if (input.cwd === f.root && ++parentCaptures === 2) vi.stubEnv("PAPERCLIP_SECRETS_MASTER_KEY", "62".repeat(32));
return result;
});
await expect(parent(f)).rejects.toThrow("Secret decryption failed");
expect(parentCaptures).toBe(2); expect([...counts.values()]).toEqual([1, 1]);
expect(readRemoteWorkspaceAuthority((await db.select().from(executionWorkspaces).where(eq(executionWorkspaces.id, f.parent.id)))[0].metadata)).toBeNull();
expect((await db.select().from(activityLog)).filter(row => row.action === "workspace.remote_storage_migrated" && row.entityId === f.parent.id)).toHaveLength(0);
}, 60_000);
it("resumes completed immutable child checks after losing the process and scratch", async () => {
const f = await imported(), get = storage.getObject.bind(storage), counts = new Map<string, number>();
storage = { ...storage, immutableObjectNamespace: storageNamespace };
vi.spyOn(storage, "getObject").mockImplementation(async (...args) => {
if (f.keys.has(args[1])) counts.set(args[1], (counts.get(args[1]) ?? 0) + 1);
return get(...args);
});
const capture = workspaceArchive.captureRemoteWorkspaceDirectory;
const interrupted = vi.spyOn(workspaceArchive, "captureRemoteWorkspaceDirectory").mockImplementation(async input => {
if (input.cwd === f.root) throw new Error("simulated_container_restart");
return capture(input);
});
await expect(parent(f)).rejects.toThrow("simulated_container_restart");
const [operation] = await db.select().from(workspaceOperations).where(eq(workspaceOperations.executionWorkspaceId, f.parent.id));
expect(operation.status).toBe("running");
expect([...counts.values()]).toEqual([1, 1]);
interrupted.mockRestore();
await expect(parent(f)).resolves.toMatchObject({ applied: true });
expect([...counts.values()]).toEqual([1, 1]);
await expect(parent(f)).resolves.toMatchObject({ applied: true, replayed: true });
const operations = await db.select().from(workspaceOperations).where(eq(workspaceOperations.executionWorkspaceId, f.parent.id));
expect(operations).toHaveLength(1); expect(operations[0].id).toBe(operation.id); expect(operations[0].status).toBe("succeeded");
const publication = await db.select().from(activityLog);
expect(publication.filter(row => row.action === "workspace.remote_storage_migrated" && row.entityId === f.parent.id)).toHaveLength(1);
}, 60_000);
it.each(["local-source", "etag", "namespace", "reference", "key", "lineage", "proof-version", "weak"] as const)(
"revalidates a changed %s after restarting from an immutable checkpoint", async change => {
const f = await imported(), get = storage.getObject.bind(storage), counts = new Map<string, number>(), snapshot = f.snapshots[0];
storage = { ...storage, immutableObjectNamespace: storageNamespace };
const reads = async (...args: Parameters<StorageService["getObject"]>) => {
if (f.keys.has(args[1])) counts.set(args[1], (counts.get(args[1]) ?? 0) + 1);
return get(...args);
};
vi.spyOn(storage, "getObject").mockImplementation(reads);
const capture = workspaceArchive.captureRemoteWorkspaceDirectory;
const stop = vi.spyOn(workspaceArchive, "captureRemoteWorkspaceDirectory").mockImplementation(async input => {
if (input.cwd === f.root) throw new Error("simulated_container_restart");
return capture(input);
});
await expect(parent(f)).rejects.toThrow("simulated_container_restart"); stop.mockRestore();
const [saved] = await db.select().from(executionWorkspaceSnapshots).where(eq(executionWorkspaceSnapshots.id, snapshot.id));
expect(saved.metadata?.nestedImportVerification).toMatchObject({ schema: "paperclip.workspace-snapshot-verification.v1" });
expect(saved.updatedAt).toEqual(snapshot.updatedAt); expect(saved.deleteAfter).toEqual(snapshot.deleteAfter);
if (change === "local-source") await fs.writeFile(path.join(f.leaves[0].cwd!, "new-work.txt"), "new work survives restart\n");
if (change === "etag") headEtags.set(snapshot.objectKey!, '"replaced-object-version"');
if (change === "weak") headEtags.set(snapshot.objectKey!, 'W/"not-a-strong-validator"');
if (change === "namespace") storage = { ...storage, immutableObjectNamespace: "97".repeat(32) };
if (change === "reference") await db.update(executionWorkspaceSnapshots).set({ sha256: "00".repeat(32) })
.where(eq(executionWorkspaceSnapshots.id, snapshot.id));
if (change === "key") await db.update(executionWorkspaceSnapshots).set({ key: { ...snapshot.key,
tag: Buffer.alloc(16).toString("base64") } }).where(eq(executionWorkspaceSnapshots.id, snapshot.id));
if (change === "lineage") await db.update(executionWorkspaceSnapshots).set({ metadata: { ...saved.metadata,
importedWorkspaceChildren: [{ workspaceId: randomUUID(), snapshotId: randomUUID(), operationId: randomUUID(),
relativePath: ".paperclip/worktrees/unproved", manifestDigest: "98".repeat(32) }] } })
.where(eq(executionWorkspaceSnapshots.id, snapshot.id));
if (change === "proof-version") await db.update(executionWorkspaceSnapshots).set({ metadata: { ...saved.metadata,
nestedImportVerification: { ...(saved.metadata!.nestedImportVerification as Record<string, unknown>), schema: "paperclip.workspace-snapshot-verification.v99" } } })
.where(eq(executionWorkspaceSnapshots.id, snapshot.id));
if (["local-source", "reference", "key", "lineage"].includes(change)) {
await expect(parent(f)).rejects.toThrow(change === "local-source" ? "nested_workspace_source_changed"
: change === "reference" ? "object_hash_mismatch" : change === "key" ? "Secret decryption failed" : "nested_workspace_coverage_invalid");
const [workspace] = await db.select().from(executionWorkspaces).where(eq(executionWorkspaces.id, f.parent.id));
expect(readRemoteWorkspaceAuthority(workspace.metadata)).toBeNull();
if (change === "local-source") {
expect([...counts.values()]).toEqual([1, 1]);
expect(await fs.readFile(path.join(f.leaves[0].cwd!, "new-work.txt"), "utf8")).toBe("new work survives restart\n");
}
} else {
await expect(parent(f)).resolves.toMatchObject({ applied: true });
expect(counts.get(snapshot.objectKey!)).toBe(change === "weak" ? 3 : 2);
if (change === "namespace") expect([...counts.values()]).toEqual([2, 2]);
}
}, 60_000);
it("overlaps two independent child readbacks and reuses only authenticated immutable proof", async () => {
const f = await imported(), get = storage.getObject.bind(storage), counts = new Map<string, number>();
const restore = vi.spyOn(workspaceArchive, "restoreRemoteWorkspaceArchive"), capture = vi.spyOn(workspaceArchive, "captureRemoteWorkspaceDirectory");
@@ -236,6 +493,7 @@ describe.skipIf(!support.supported)("bounded nested workspace verification", ()
}, 60_000);
it("shares immutable proof across declared aliases inside one complete source fence", async () => {
const f = await imported(), get = storage.getObject.bind(storage), counts = new Map<string, number>();
storage = { ...storage, immutableObjectNamespace: storageNamespace };
const [issue] = await db.insert(issues).values({ companyId: f.companyId, projectId: f.projectId, title: "Parent alias" }).returning();
const [alias] = await db.insert(executionWorkspaces).values({ companyId: f.companyId, projectId: f.projectId,
sourceIssueId: issue.id, name: "Parent alias", mode: "shared_workspace", strategyType: "project_primary", providerType: "local_fs",
@@ -259,5 +517,6 @@ describe.skipIf(!support.supported)("bounded nested workspace verification", ()
} finally { await verifier.close(); }
});
expect([...counts.values()]).toEqual([1, 1]);
expect((await db.select().from(executionWorkspaceSnapshots)).every(row => !row.metadata?.nestedImportVerification)).toBe(true);
}, 60_000);
});
@@ -360,6 +360,10 @@ async function snapshotKey(ref: WorkspaceSnapshotRef) {
return wrapped;
}
export async function assertWorkspaceSnapshotKeyReadable(reference: WorkspaceSnapshotRef): Promise<void> {
await snapshotKey(reference);
}
/** Authentication finishes in private scratch before any plaintext is published. */
export async function loadWorkspaceSnapshotArchive(input: {
storage: StorageService; reference: WorkspaceSnapshotRef; targetPath: string; signal?: AbortSignal; temporaryParent?: string;
@@ -0,0 +1,83 @@
import { createHash } from "node:crypto";
import { isDeepStrictEqual } from "node:util";
import { and, eq, sql } from "drizzle-orm";
import { executionWorkspaceSnapshots, type Db } from "@paperclipai/db";
import type { ImportedWorkspaceChild } from "./remote-workspace-migration-tree.js";
import type { WorkspaceSnapshotRef } from "./remote-workspace-snapshot-storage.js";
const SCHEMA = "paperclip.workspace-snapshot-verification.v1";
const KEY = "nestedImportVerification";
type Snapshot = typeof executionWorkspaceSnapshots.$inferSelect;
type Binding = { reference: WorkspaceSnapshotRef; children: ImportedWorkspaceChild[]; operationId: string; etag: string };
function canonical(value: unknown): string {
if (Array.isArray(value)) return `[${value.map(canonical).join(",")}]`;
if (value && typeof value === "object") return `{${Object.entries(value).sort(([a], [b]) => a.localeCompare(b))
.map(([key, item]) => `${JSON.stringify(key)}:${canonical(item)}`).join(",")}}`;
return JSON.stringify(value) ?? "null";
}
function fingerprint(namespace: string, input: Binding) {
return createHash("sha256").update(canonical([SCHEMA, namespace, input.reference, input.children, input.operationId])).digest("hex");
}
function matches(row: Snapshot | undefined, input: Binding) {
const ref = input.reference;
return row && ["committed", "superseded"].includes(row.status) && row.sourceOperationId === input.operationId
&& row.companyId === ref.companyId && row.executionWorkspaceId === ref.workspaceId && row.id === ref.id
&& row.generation === ref.generation && row.objectKey === ref.objectKey && row.sha256 === ref.sha256
&& row.byteSize === ref.byteSize && row.manifestDigest === ref.manifestDigest && isDeepStrictEqual(row.key, ref.key)
&& isDeepStrictEqual(row.metadata?.importedWorkspaceChildren ?? [], input.children);
}
function scope(ref: WorkspaceSnapshotRef) {
return and(eq(executionWorkspaceSnapshots.companyId, ref.companyId), eq(executionWorkspaceSnapshots.executionWorkspaceId, ref.workspaceId),
eq(executionWorkspaceSnapshots.id, ref.id));
}
function transientCheckpointFailure(error: unknown): boolean {
if (!error || typeof error !== "object") return false;
const failure = error as { code?: unknown; cause?: { code?: unknown } };
const code = typeof failure.code === "string" ? failure.code : failure.cause?.code;
return typeof code === "string" && ["55P03", "57014", "40P01", "40001"].includes(code);
}
export function nestedWorkspaceVerificationCheckpoint(db: Db, namespace: string | undefined, signal?: AbortSignal) {
if (!namespace || !/^[a-f0-9]{64}$/.test(namespace)) return undefined;
return {
async read(input: Binding): Promise<boolean> {
signal?.throwIfAborted();
try { return await db.transaction(async tx => {
await tx.execute(sql`set local lock_timeout='1s'`);
await tx.execute(sql`set local statement_timeout='8s'`);
const [row] = await tx.select().from(executionWorkspaceSnapshots).where(scope(input.reference));
signal?.throwIfAborted();
if (!matches(row, input)) return false;
const raw = row.metadata?.[KEY];
if (!raw || typeof raw !== "object" || Array.isArray(raw)) return false;
const proof = raw as Record<string, unknown>;
return Object.keys(proof).every(key => ["schema", "namespace", "fingerprint", "etag", "completedAt"].includes(key))
&& proof.schema === SCHEMA && proof.namespace === namespace && proof.fingerprint === fingerprint(namespace, input)
&& proof.etag === input.etag && typeof proof.completedAt === "string" && Number.isFinite(Date.parse(proof.completedAt));
}, { accessMode: "read only" }); } catch (error) {
signal?.throwIfAborted();
if (!transientCheckpointFailure(error)) throw error;
return false;
}
},
async save(input: Binding): Promise<void> {
signal?.throwIfAborted();
try { await db.transaction(async tx => {
await tx.execute(sql`set local lock_timeout='1s'`);
await tx.execute(sql`set local statement_timeout='8s'`);
const [row] = await tx.select().from(executionWorkspaceSnapshots).where(scope(input.reference)).for("update");
signal?.throwIfAborted();
if (!matches(row, input)) throw new Error("nested_workspace_snapshot_changed");
const proof = { schema: SCHEMA, namespace, fingerprint: fingerprint(namespace, input), etag: input.etag,
completedAt: new Date().toISOString() };
await tx.update(executionWorkspaceSnapshots).set({ metadata: { ...row.metadata, [KEY]: proof } }).where(scope(input.reference));
signal?.throwIfAborted();
}); } catch (error) {
signal?.throwIfAborted();
if (!transientCheckpointFailure(error)) throw error;
}
},
};
}
+4
View File
@@ -9,6 +9,7 @@ import {
} from "@aws-sdk/client-s3";
import { assertSourceFileSize, fileRangeMd5, putS3Multipart, putS3MultipartFile } from "./s3-multipart.js";
import { createReadStream } from "node:fs";
import { createHash } from "node:crypto";
import { recordS3BodyFailure, withS3FailureDiagnostics } from "./s3-failure.js";
import { addAbortSignal, Readable } from "node:stream";
import { retryS3RequestTimeouts } from "./s3-request-timeout-retry.js";
@@ -121,6 +122,9 @@ export function createS3StorageProvider(config: S3ProviderConfig): StorageProvid
return {
id: "s3",
immutableObjectNamespace: config.endpoint ? createHash("sha256").update(JSON.stringify(["s3",
new URL(config.endpoint).href, region, bucket, prefix, Boolean(config.forcePathStyle),
])).digest("hex") : undefined,
async putObjectConditional(input) {
const output = await client.send(new PutObjectCommand({
+1
View File
@@ -107,6 +107,7 @@ function assertPutFileInput(input: PutFileInput): void {
export function createStorageService(provider: StorageProvider): StorageService {
return {
provider: provider.id,
...(provider.immutableObjectNamespace ? { immutableObjectNamespace: provider.immutableObjectNamespace } : {}),
async putFile(input: PutFileInput): Promise<PutFileResult> {
assertPutFileInput(input);
+2
View File
@@ -61,6 +61,7 @@ export interface HeadObjectResult {
export interface StorageProvider {
id: StorageProviderId;
readonly immutableObjectNamespace?: string;
putObject(input: PutObjectInput): Promise<void | PutObjectReceipt>;
getObject(input: GetObjectInput): Promise<GetObjectResult>;
headObject(input: GetObjectInput): Promise<HeadObjectResult>;
@@ -100,6 +101,7 @@ export interface PutFileResult {
export interface StorageService {
provider: StorageProviderId;
readonly immutableObjectNamespace?: string;
putFile(input: PutFileInput): Promise<PutFileResult>;
getObject(companyId: string, objectKey: string, options?: Pick<GetObjectInput, "range" | "signal">): Promise<GetObjectResult>;
headObject(companyId: string, objectKey: string, options?: Pick<GetObjectInput, "signal">): Promise<HeadObjectResult>;