Merge pull request #29321 from overleaf/bg-ignore-deleted-projects-in-history-backup

handle deleted projects in history-v1 backup worker

GitOrigin-RevId: f4392045074248137f15d082d922c18b1ef9232f

services/history-v1/storage/lib/backup_store/index.js

@@ -1,5 +1,5 @@
 const { Binary, ObjectId } = require('mongodb')
-const { projects, backedUpBlobs } = require('../mongodb')
+const { projects, deletedProjects, backedUpBlobs } = require('../mongodb')
 const OError = require('@overleaf/o-error')
 
 // List projects with pending backups older than the specified interval
@@ -79,6 +79,13 @@ async function getBackupStatus(projectId) {
     }
   )
   if (!project) {
+    // Check whether the project was deleted
+    const deletedProject = await deletedProjects.findOne({
+      'deleterData.deletedProjectId': new ObjectId(projectId),
+    })
+    if (deletedProject) {
+      throw new Error('Project deleted')
+    }
     throw new Error('Project not found')
   }
   return {

services/history-v1/storage/lib/mongodb.js

@@ -14,6 +14,7 @@ const blobs = db.collection('projectHistoryBlobs')
 const globalBlobs = db.collection('projectHistoryGlobalBlobs')
 const shardedBlobs = db.collection('projectHistoryShardedBlobs')
 const projects = db.collection('projects')
+const deletedProjects = db.collection('deletedProjects')
 // Temporary collection for tracking progress of backed up old blobs (without a hash).
 // The initial sync process will be able to skip over these.
 // Schema: _id: projectId, blobs: [Binary]
@@ -32,6 +33,7 @@ module.exports = {
   blobs,
   globalBlobs,
   projects,
+  deletedProjects,
   shardedBlobs,
   backedUpBlobs,
   cleanupTestDatabase,

services/history-v1/storage/scripts/backup_worker.mjs

@@ -114,9 +114,14 @@ async function runBackup(projectId, data, job) {
     }
     return `backup completed ${projectId}`
   } catch (err) {
-    metrics.inc('backup_worker_project', 1, { status: 'failed' })
-    logger.error({ projectId, err }, 'backup failed')
-    throw err // Re-throw to mark job as failed
+    if (err.message === 'Project deleted') {
+      metrics.inc('backup_worker_project', 1, { status: 'deleted' })
+      logger.warn({ projectId, err }, 'skipping backup of deleted project')
+    } else {
+      metrics.inc('backup_worker_project', 1, { status: 'failed' })
+      logger.error({ projectId, err }, 'backup failed')
+      throw err // Re-throw to mark job as failed
+    }
   }
 }