Fix repository cleanup after removal and expiration (#782)

This commit is contained in:
Thomas Durieux
2026-08-20 12:52:19 +02:00
committed by GitHub
parent 46e956a779
commit 35c3b1804b
8 changed files with 370 additions and 69 deletions
+12 -1
View File
@@ -17,9 +17,14 @@ import router from "./routes";
import {
conferenceStatusCheck,
repositoryStatusCheck,
runRepositoryStatusCheck,
dailyStatsSnapshot,
} from "./schedule";
import { startWorker, recoverStuckPreparing } from "../queue";
import {
startWorker,
recoverStuckPreparing,
recoverStuckRemoving,
} from "../queue";
import {
computeStats,
ensureTodaySnapshot,
@@ -363,6 +368,12 @@ export default async function start() {
recoverStuckPreparing().catch((err) =>
logger.error("recoverStuckPreparing failed", serializeError(err))
);
recoverStuckRemoving().catch((err) =>
logger.error("recoverStuckRemoving failed", serializeError(err))
);
runRepositoryStatusCheck().catch((err) =>
logger.error("initial repository status check failed", serializeError(err))
);
}
start();
+9 -2
View File
@@ -17,7 +17,7 @@ import { IAnonymizedRepositoryDocument } from "../../core/model/anonymizedReposi
import UserModel from "../../core/model/users/users.model";
import ConferenceModel from "../../core/model/conference/conferences.model";
import AnonymousError from "../../core/AnonymousError";
import { downloadQueue, removeQueue } from "../../queue";
import { addRemovalJob, downloadQueue } from "../../queue";
import RepositoryModel from "../../core/model/repositories/repositories.model";
import User from "../../core/User";
import { RepositoryStatus } from "../../core/types";
@@ -241,7 +241,14 @@ router.delete(
const user = await getUser(req);
isOwnerOrAdmin([repo.owner.id], user);
await repo.updateStatus(RepositoryStatus.REMOVING);
await removeQueue.add(repo.repoId, { repoId: repo.repoId }, { jobId: `repo-${repo.repoId}` });
try {
await addRemovalJob(repo.repoId);
} catch (error) {
const message =
error instanceof Error ? error.message : "removal_enqueue_failed";
await repo.updateStatus(RepositoryStatus.ERROR, message);
throw error;
}
return res.json({ status: repo.status });
} catch (error) {
handleError(error, res, req);
+109 -44
View File
@@ -30,55 +30,120 @@ export function conferenceStatusCheck() {
export function repositoryStatusCheck() {
// check every 6 hours the status of the repositories
schedule.scheduleJob("0 */6 * * *", async () => {
logger.info("checking repository status and unused repositories");
const now = new Date();
const fourMonthAgo = new Date(now);
fourMonthAgo.setMonth(fourMonthAgo.getMonth() - 4);
const cursor = AnonymizedRepositoryModel.find({
status: RepositoryStatus.READY,
isReseted: false,
$or: [
{
"options.expirationMode": { $in: ["redirect", "remove"] },
"options.expirationDate": { $lte: now },
},
{ lastView: { $lt: fourMonthAgo } },
],
}).cursor();
const batch: Promise<void>[] = [];
for await (const data of cursor) {
batch.push(
(async () => {
const repo = new Repository(data);
try {
await repo.check();
} catch {
logger.info("repository expired", { repoId: repo.repoId });
}
await runRepositoryStatusCheck();
});
}
if (repo.model.lastView < fourMonthAgo) {
try {
await repo.removeCache();
} catch (error) {
logger.error("repository cache removal failed", {
...serializeError(error),
repoId: repo.repoId,
});
return;
}
logger.info("removed cache for unused repository", {
export function repositoryMaintenanceQuery(now: Date) {
const fourMonthAgo = new Date(now);
fourMonthAgo.setMonth(fourMonthAgo.getMonth() - 4);
return {
status: RepositoryStatus.READY,
$or: [
{
"options.expirationMode": { $in: ["redirect", "remove"] },
"options.expirationDate": { $lte: now },
},
{
isReseted: { $ne: true },
lastView: { $lt: fourMonthAgo },
},
],
};
}
export async function runRepositoryStatusCheck(now = new Date()) {
logger.info("checking repository status and unused repositories");
const fourMonthAgo = new Date(now);
fourMonthAgo.setMonth(fourMonthAgo.getMonth() - 4);
const batch: Promise<void>[] = [];
const flushBatch = async () => {
await Promise.all(batch);
batch.length = 0;
};
const cursor = AnonymizedRepositoryModel.find(
repositoryMaintenanceQuery(now)
).cursor();
for await (const data of cursor) {
batch.push(
(async () => {
const repo = new Repository(data);
const shouldExpire =
repo.options.expirationMode !== "never" &&
repo.options.expirationDate != null &&
repo.options.expirationDate <= now;
if (shouldExpire) {
try {
await repo.expire();
logger.info("repository expired", { repoId: repo.repoId });
} catch (error) {
logger.error("repository expiration failed", {
...serializeError(error),
repoId: repo.repoId,
});
}
})()
);
if (batch.length >= 10) {
await Promise.all(batch);
batch.length = 0;
}
return;
}
if (
repo.model.isReseted !== true &&
repo.model.lastView < fourMonthAgo
) {
try {
await repo.removeCache();
logger.info("removed cache for unused repository", {
repoId: repo.repoId,
});
} catch (error) {
logger.error("repository cache removal failed", {
...serializeError(error),
repoId: repo.repoId,
});
}
}
})()
);
if (batch.length >= 10) {
await flushBatch();
}
await Promise.all(batch);
});
}
await flushBatch();
// Repair terminal records left with data by an older or interrupted
// expiration. This makes the cleanup idempotent across deployments.
const dirtyTerminalCursor = AnonymizedRepositoryModel.find({
$or: [
{ status: RepositoryStatus.EXPIRING },
{
status: RepositoryStatus.EXPIRED,
isReseted: { $ne: true },
},
],
}).cursor();
for await (const data of dirtyTerminalCursor) {
batch.push(
(async () => {
const repo = new Repository(data);
try {
await repo.resetSate();
await repo.updateStatus(RepositoryStatus.EXPIRED);
logger.info("recovered expired repository cleanup", {
repoId: repo.repoId,
});
} catch (error) {
logger.error("expired repository cleanup failed", {
...serializeError(error),
repoId: repo.repoId,
});
}
})()
);
if (batch.length >= 10) {
await flushBatch();
}
}
await flushBatch();
}
export function dailyStatsSnapshot() {