trails/apps/journal/app/lib/federation-queue.server.ts
Ullrich Schäfer b8fc8fadff
feat(journal): durable Fedify message queue over pg-boss
Task group 1 of federation-hardening. Fedify was configured with
`InProcessMessageQueue`, so every queued outbound delivery, its pending
retry state, and inbox processing task was lost on a container restart
(routine here: deploys, OOM history) — directly contradicting the
social-federation spec's promise that fan-out survives a deploy.

- `federation-queue.server.ts`: `PgBossMessageQueue` implementing Fedify's
  `MessageQueue` over the pg-boss instance the journal already runs,
  mirroring the `PostgresKvStore` adapter. `nativeRetrial = false` keeps
  Fedify the retry-policy owner; pg-boss supplies durability + delayed
  jobs (delay → whole-second `startAfter`, `retryLimit: 0`).
- Swap it in for `InProcessMessageQueue` in `federation.server.ts`;
  document the two intentional queueing layers (our fan-out jobs feed
  Fedify; Fedify's sends now durable underneath).
- `server.ts`: create the durable queue at startup when federation is on.
- Broaden the structural `BossLike` in `boss.server.ts` with the
  work/offWork/createQueue/getQueue methods the adapter needs.
- Tests: enqueue maps retry/delay correctly, consume roundtrip, restart
  durability (fresh listener drains a prior instance's backlog), abort
  stops the worker, depth reports ready vs delayed.

Verified: journal typecheck + lint clean, 7/7 new unit tests pass.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-13 21:46:46 +02:00

77 lines
3.3 KiB
TypeScript

// Postgres-backed Fedify MessageQueue, implemented over the pg-boss
// instance the journal already runs (packages/jobs). Fedify uses this
// queue for outbound delivery (with its own retry scheduling) and inbox
// processing; the previous InProcessMessageQueue lost every queued
// message on restart (spec: federation-operations "Durable federation
// queue"). pg-boss gives us durable storage + delayed jobs; Fedify stays
// the retry-policy owner (see `nativeRetrial` below).
//
// This mirrors the PostgresKvStore adapter pattern (federation-kv.server.ts):
// a thin Fedify-interface shim over infrastructure we already operate,
// resolved lazily via getBoss() so it works in both the server-bootstrap
// module and the request-path bundle (see boss.server.ts on the two-copy
// module situation).
//
// Fedify fans a single MessageQueue instance out to three logical roles
// (inbox / outbox / fanout) and calls listen() once per role with an
// equivalent type-dispatching handler. Backing all three with one pg-boss
// queue is correct: any worker can process any message because Fedify
// routes by message type internally.
import type {
MessageQueue,
MessageQueueEnqueueOptions,
MessageQueueListenOptions,
MessageQueueDepth,
} from "@fedify/fedify";
import { getBoss } from "./boss.server.ts";
/** pg-boss queue name backing Fedify's inbox/outbox/fanout message queue. */
export const FEDERATION_QUEUE_NAME = "federation.fedify";
export class PgBossMessageQueue implements MessageQueue {
// Fedify owns retry scheduling: on a failed delivery it re-enqueues with
// its own backoff policy, so pg-boss must NOT also retry (retryLimit 0
// on enqueue) or every failure would be attempted twice.
readonly nativeRetrial = false;
async enqueue(message: unknown, options?: MessageQueueEnqueueOptions): Promise<void> {
// Fedify passes delay as a Temporal.Duration; pg-boss startAfter is
// whole seconds. Round up so a sub-second backoff still defers.
const seconds = options?.delay ? Math.ceil(options.delay.total("seconds")) : 0;
await getBoss().send(FEDERATION_QUEUE_NAME, message as object, {
retryLimit: 0,
...(seconds > 0 ? { startAfter: seconds } : {}),
});
}
async listen(
handler: (message: unknown) => Promise<void> | void,
options?: MessageQueueListenOptions,
): Promise<void> {
const boss = getBoss();
await boss.work(FEDERATION_QUEUE_NAME, async (jobs) => {
for (const job of jobs) await handler(job.data);
});
// Fedify's contract: listen() resolves when the signal aborts and never
// rejects; with no signal it never resolves (runs for the process
// lifetime, consuming the queue).
return new Promise<void>((resolve) => {
const signal = options?.signal;
if (signal == null) return;
const stop = () => void boss.offWork(FEDERATION_QUEUE_NAME).finally(() => resolve());
if (signal.aborted) stop();
else signal.addEventListener("abort", stop, { once: true });
});
}
async getDepth(): Promise<MessageQueueDepth> {
const queue = await getBoss().getQueue(FEDERATION_QUEUE_NAME);
if (queue == null) return { queued: 0, ready: 0, delayed: 0 };
return {
queued: queue.queuedCount,
ready: queue.readyCount,
delayed: queue.deferredCount,
};
}
}