perf(storage): upload workspace snapshots in parallel verified parts (#19917)

## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - With remote (sandbox) execution, every agent turn ends by saving the
workspace: capture an archive, encrypt it, upload it to object storage,
and verify it.
> - On a busy S3-backed instance, the upload step is the slowest part of
a save. A 1.5 GB workspace can hold a run open for minutes after the
agent finished.
> - The upload step reads the archive several times, uploads through one
stream with 4 parts in flight, and then downloads the whole object again
to verify it.
> - Parts that use the stream path share one process-wide 32 MiB buffer
window, so saves that run at the same time slow each other down.
> - This pull request uploads parts in parallel straight from the
encrypted file, lets the backend verify every part with Content-MD5, and
replaces the full read-back with a size check for those verified
uploads.
> - The benefit is faster saves, a smaller run-finalization window, and
no interference between concurrent saves.

## Linked Issues or Issue Description

**What existing behavior does this improve?**
Remote workspace snapshot upload (`uploadWorkspaceSnapshot`) to
S3-compatible storage.

**Subsystem affected**
Server storage (`server/src/storage/*`) and remote workspace snapshot
storage.

**Current behavior**
For one save, the upload step:
1. hashes the archive;
2. reads the archive again to encrypt it into a temporary file;
3. hashes the encrypted file;
4. uploads it as one stream with 4 parts in flight;
5. downloads the whole object to verify it.

On one hosted instance over 7 days (724 saves, median 731 MB), the
upload step ran at a median of 6.5 MB/s. With no other save running it
reached 11–14 MB/s. With six or more saves at the same time it fell to
4.4 MB/s. From the same region and with the same part settings, a plain
client reached 31 MB/s to the same kind of bucket.

**Proposed behavior**
Upload parts in parallel from byte ranges of the encrypted file, with
Content-MD5 on each part. Confirm only the stored size after a verified
upload. Digest the archive in the same pass that encrypts it.

**Reason and benefit**
Saves finish faster and hold their run open for less time. Concurrent
saves no longer serialize on the shared buffer window. A transient part
failure is retried from disk instead of failing the whole save.

**Breaking changes**
None. `StorageProvider.putObject` may now return a receipt. Providers
that return nothing keep the current behavior, including the full
read-back.

Related: none found.

## What Changed

- `storage/types.ts`, `storage/service.ts`: an upload can name its
`sourceFile`, a private file whose bytes are exactly the body.
`putObject` may return `{ integrity: "server_verified_md5" }`, and
`putFile` passes it through as `PutFileResult.integrity`.
- `storage/s3-multipart.ts`: new `putS3MultipartFile`.
- Each part reads its own byte range of the file. Up to 8 parts upload
at once, with no shared buffer window.
  - Each part carries the Content-MD5 of its range.
  - A failed part is resent from disk, up to 3 attempts.
- The first part that runs out of attempts stops its siblings. The
multipart upload is aborted only after every part settles.
- A source file whose size differs from the declared length is refused
before any request.
- `storage/s3-provider.ts`:
- A `sourceFile` larger than 16 MiB uses `putS3MultipartFile`. A smaller
one uses one PUT with Content-MD5.
- These requests use a second client with `requestChecksumCalculation:
"WHEN_REQUIRED"`, so bodies are plain and the backend verifies
Content-MD5 directly. The default client wraps bodies in aws-chunked
framing with a CRC32 trailer.
  - Other uploads are unchanged.
- `services/remote-workspace-snapshot-storage.ts`:
- Encryption digests the plaintext in the same read. An unchanged file
identity (dev, inode, size, mtime, ctime) across that read replaces the
separate digest pass. A change still fails with `archive_changed` before
any upload.
- The encrypted file is still digested from disk, so the recorded hash
covers the exact bytes that the parts read.
- The upload passes `sourceFile`. When the provider reports
`server_verified_md5`, a HEAD confirms the stored size instead of
downloading the object.
  - The full read-back stays for:
    - other providers (local disk);
    - recovery of earlier attempts;
    - the caller that keeps a verified archive file.

## Verification

- `pnpm --filter @paperclipai/server typecheck`: exit 0.
- `vitest run src/__tests__/storage-s3-multipart.test.ts
src/services/remote-workspace-snapshot-storage.test.ts`: 59 passed. New
tests:
- every part carries the Content-MD5 of its range, and 8 parts are in
flight;
  - a failed part is resent and the upload completes;
- a part that keeps failing aborts the upload only after in-flight
siblings settle;
  - a size mismatch is refused before any request;
- a small file goes as one PUT with Content-MD5, through a real HTTP
server;
- the read-back is skipped only for a server-verified upload, and the
snapshot still restores;
  - the read-back stays when the caller keeps the read-back file;
  - a stored size mismatch is rejected;
  - an archive that changes during encryption is rejected before upload.
- With the source change reverted, the three new snapshot tests fail.
- Real S3-compatible bucket (a Railway bucket, run from a workstation):
- A 3 MiB single PUT stored exactly: HEAD size, ETag and read-back
SHA-256 all matched.
  - A 96 MiB, 13-part upload stored exactly, with the same checks.
  - A part with a wrong Content-MD5 was rejected with `BadDigest`.
- The same backend stored a wrong `x-amz-checksum-sha256` without
checking it. That is why this change uses Content-MD5.
- Measured from a container in the app's region, 1 GiB with the app's
settings (8 MiB parts, 4 in flight), to the same kind of bucket: 31
MB/s. With 16 in flight: 69 MB/s.
- 77 storage-related suites: the only failures are the same 19 that fail
on unmodified `master` in this checkout. They are in 4 files: missing
bundled connector skills and an unbuilt local runner daemon.

## Risks

- Integrity now depends on the backend enforcing Content-MD5 for file
uploads. AWS S3 and the tested S3-compatible backend enforce it. A
backend that ignores Content-MD5 would accept corrupted parts without a
read-back. A later restore still fails closed: AES-GCM authentication
and the plaintext SHA-256 are checked on load.
- Up to 8 parallel part requests per save, instead of 4 shared ones.
Memory per part is a 64 KiB stream buffer.
- Archive-change detection now uses file identity instead of a second
digest. The archive lives in a private temporary directory that the
capture step creates.
- Medium risk overall: the change is limited to file-sourced S3 uploads
and the snapshot verify step.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- Claude Opus 5.5 (`claude-opus-5-5`) in Claude Code, with tool use
(shell, file edits, test runs, S3 API probes). AI-assisted
investigation, change and tests, with human direction.

## 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
docs describe this path)
- [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

Co-authored-by: Paperclip <noreply@paperclip.ing>
This commit is contained in:
Devin FoleyandPaperclip authored and GitHub committed 2026-11-07 14:04:49 -08:00
1 parent 99529f9d71
commit ee556a266d
7 files changed
+385 -17

No files matched your search

@@ -4,7 +4,10 @@ import { createHash } from "node:crypto";
import { createServer, type ServerResponse } from "node:http";
import type { AddressInfo } from "node:net";
import { describe, expect, it, vi } from "vitest";
import { putS3Multipart } from "../storage/s3-multipart.js";
import { FILE_PART_CONCURRENCY, putS3Multipart, putS3MultipartFile } from "../storage/s3-multipart.js";
import fs from "node:fs/promises";
import os from "node:os";
import path from "node:path";
import { createS3StorageProvider } from "../storage/s3-provider.js";
import { createStorageService } from "../storage/service.js";
import { STORAGE_FAILURE_CLEANUP_MS } from "../storage/failure-cleanup.js";
@@ -419,3 +422,157 @@ describe("streamed S3 multipart uploads", () => {
}
});
});
describe("file-backed S3 multipart uploads", () => {
const partBytes = 8 * 1024 * 1024;
async function sourceFile(parts: number, tail = 0) {
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "pc-s3-file-part-"));
const file = path.join(dir, "source.bin");
const handle = await fs.open(file, "wx", 0o600);
try {
for (let index = 0; index < parts; index++) await handle.write(Buffer.alloc(partBytes, index + 1));
if (tail) await handle.write(Buffer.alloc(tail, 0xee));
} finally { await handle.close(); }
return { dir, file, size: parts * partBytes + tail };
}
const md5 = (bytes: Buffer) => createHash("md5").update(bytes).digest("base64");
async function drain(body: unknown) {
const chunks: Buffer[] = [];
for await (const chunk of body as Readable) chunks.push(Buffer.from(chunk));
return Buffer.concat(chunks);
}
it("uploads ranges in parallel and sends each part's Content-MD5", async () => {
const source = await sourceFile(10, 4096);
const client = new S3Client({ region: "us-east-1" });
let active = 0, peak = 0, release!: () => void;
const gate = new Promise<void>(resolve => { release = resolve; });
const seen = new Map<number, { length: number; md5: string; declared: string | undefined }>();
const send = vi.spyOn(client, "send").mockImplementation(async (command: any) => {
if (command instanceof CreateMultipartUploadCommand) return { UploadId: "upload" };
if (command instanceof UploadPartCommand) {
active++; peak = Math.max(peak, active);
if (active === FILE_PART_CONCURRENCY) release();
await gate;
const bytes = await drain(command.input.Body);
expect(bytes.length).toBe(command.input.ContentLength);
seen.set(command.input.PartNumber!, { length: bytes.length, md5: md5(bytes), declared: command.input.ContentMD5 });
active--;
return { ETag: `part-${command.input.PartNumber}` };
}
expect(command).toBeInstanceOf(CompleteMultipartUploadCommand);
expect(command.input.MultipartUpload.Parts.map((part: { PartNumber: number }) => part.PartNumber)).toEqual([...Array(11).keys()].map(n => n + 1));
return {};
});
const body = Readable.from([Buffer.from("unused stream body")]);
try {
const receipt = await putS3MultipartFile(client, "bucket", "key", { objectKey: "key", contentType: "application/octet-stream",
contentLength: source.size, body, sourceFile: source.file });
expect(receipt).toEqual({ integrity: "server_verified_md5" });
expect(peak).toBe(FILE_PART_CONCURRENCY);
expect(body.destroyed).toBe(true);
expect(seen.size).toBe(11);
for (const [number, part] of seen) {
expect(part.declared).toBe(part.md5);
expect(part.length).toBe(number === 11 ? 4096 : partBytes);
}
expect(seen.get(3)!.md5).toBe(md5(Buffer.alloc(partBytes, 3)));
} finally { send.mockRestore(); client.destroy(); await fs.rm(source.dir, { recursive: true, force: true }); }
});
it("resends a failed part from the file instead of failing the upload", async () => {
const source = await sourceFile(3);
const client = new S3Client({ region: "us-east-1" });
const attempts = new Map<number, number>();
const send = vi.spyOn(client, "send").mockImplementation(async (command: any) => {
if (command instanceof CreateMultipartUploadCommand) return { UploadId: "upload" };
if (command instanceof UploadPartCommand) {
const number = command.input.PartNumber!;
attempts.set(number, (attempts.get(number) ?? 0) + 1);
const bytes = await drain(command.input.Body);
if (number === 2 && attempts.get(number) === 1) throw new Error("An error was encountered in a non-retryable streaming request.");
expect(md5(bytes)).toBe(command.input.ContentMD5);
return { ETag: `part-${number}` };
}
if (command instanceof AbortMultipartUploadCommand) throw new Error("must not abort a recovered upload");
return {};
});
try {
await expect(putS3MultipartFile(client, "bucket", "key", { objectKey: "key", contentType: "application/octet-stream",
contentLength: source.size, body: Buffer.alloc(0), sourceFile: source.file })).resolves.toEqual({ integrity: "server_verified_md5" });
expect(attempts.get(2)).toBe(2);
expect(send.mock.calls.some(([command]) => command instanceof CompleteMultipartUploadCommand)).toBe(true);
} finally { send.mockRestore(); client.destroy(); await fs.rm(source.dir, { recursive: true, force: true }); }
});
it("aborts the upload only after every in-flight part settles when a part keeps failing", async () => {
const source = await sourceFile(4);
const client = new S3Client({ region: "us-east-1" });
let inFlight = 0, abortedWhileInFlight = false;
const send = vi.spyOn(client, "send").mockImplementation(async (command: any, options?: any) => {
if (command instanceof CreateMultipartUploadCommand) return { UploadId: "upload" };
if (command instanceof UploadPartCommand) {
inFlight++;
try {
if (command.input.PartNumber === 1) throw new Error("storage refused part");
await new Promise<void>((_resolve, reject) => options?.abortSignal?.addEventListener("abort", () => reject(options.abortSignal!.reason), { once: true }));
return { ETag: "never" };
} finally { inFlight--; }
}
if (command instanceof AbortMultipartUploadCommand) { abortedWhileInFlight = inFlight > 0; return {}; }
throw new Error("must not complete a failed upload");
});
try {
await expect(putS3MultipartFile(client, "bucket", "key", { objectKey: "key", contentType: "application/octet-stream",
contentLength: source.size, body: Buffer.alloc(0), sourceFile: source.file })).rejects.toThrow("storage refused part");
const partOneAttempts = send.mock.calls.filter(([command]) => command instanceof UploadPartCommand && command.input.PartNumber === 1).length;
expect(partOneAttempts).toBe(3);
expect(send.mock.calls.filter(([command]) => command instanceof AbortMultipartUploadCommand)).toHaveLength(1);
expect(abortedWhileInFlight).toBe(false);
} finally { send.mockRestore(); client.destroy(); await fs.rm(source.dir, { recursive: true, force: true }); }
});
it("refuses a source file whose size differs from the declared length before starting an upload", async () => {
const source = await sourceFile(3);
const client = new S3Client({ region: "us-east-1" });
const send = vi.spyOn(client, "send");
try {
await expect(putS3MultipartFile(client, "bucket", "key", { objectKey: "key", contentType: "application/octet-stream",
contentLength: source.size + 1, body: Buffer.alloc(0), sourceFile: source.file })).rejects.toThrow("does not match its declared size");
expect(send).not.toHaveBeenCalled();
} finally { send.mockRestore(); client.destroy(); await fs.rm(source.dir, { recursive: true, force: true }); }
});
it("sends a small file source as one PUT with Content-MD5 through the real S3 HTTP client", async () => {
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "pc-s3-small-file-"));
const file = path.join(dir, "small.bin"), bytes = Buffer.from("verified snapshot bytes\n".repeat(1000));
await fs.writeFile(file, bytes, { mode: 0o600 });
const requests: Array<{ method: string; md5: string | undefined; body: Buffer }> = [];
const server = createServer((request, response) => {
const chunks: Buffer[] = [];
request.on("data", chunk => chunks.push(chunk));
request.on("end", () => {
requests.push({ method: request.method!, md5: request.headers["content-md5"] as string | undefined, body: Buffer.concat(chunks) });
response.writeHead(200, { ETag: '"small"' }).end();
});
});
await new Promise<void>((resolve, reject) => { server.once("error", reject); server.listen(0, "127.0.0.1", resolve); });
vi.stubEnv("AWS_ACCESS_KEY_ID", "fixture"); vi.stubEnv("AWS_SECRET_ACCESS_KEY", "fixture");
const service = createStorageService(createS3StorageProvider({ bucket: "bucket", region: "us-east-1", forcePathStyle: true,
endpoint: `http://127.0.0.1:${(server.address() as AddressInfo).port}` }));
try {
const result = await service.putFile({ companyId: "company", objectKey: "company/workspace-snapshots/small.pcws", namespace: "workspace-snapshots",
originalFilename: null, contentType: "application/octet-stream", body: Readable.from([bytes]), byteSize: bytes.length,
sha256: createHash("sha256").update(bytes).digest("hex"), sourceFile: file });
expect(result.integrity).toBe("server_verified_md5");
expect(requests).toHaveLength(1);
expect(requests[0]).toMatchObject({ method: "PUT", md5: md5(bytes) });
expect(requests[0]!.body.equals(bytes)).toBe(true);
} finally {
vi.unstubAllEnvs();
server.closeAllConnections();
await new Promise<void>(resolve => server.close(() => resolve()));
await fs.rm(dir, { recursive: true, force: true });
}
});
});