Add a script for finding projects missing a history id in Redis

GitOrigin-RevId: 3606e82c8e0fc8e1118cd6ce1981610ec20b73af
2026-11-14 22:23:42 +01:00 · 2025-04-29 08:05:27 +00:00
parent c2026d34b1
commit 65a57001ef
1 changed files with 199 additions and 0 deletions
@@ -0,0 +1,199 @@
+// @ts-check
+
+const Settings = require('@overleaf/settings')
+const logger = require('@overleaf/logger')
+const RedisManager = require('../app/js/RedisManager')
+const minimist = require('minimist')
+const { db, ObjectId } = require('../app/js/mongodb')
+const ProjectManager = require('../app/js/ProjectManager')
+const OError = require('@overleaf/o-error')
+
+const docUpdaterKeys = Settings.redis.documentupdater.key_schema
+
+const rclient = RedisManager.rclient
+
+const { verbose, commit, ...args } = minimist(process.argv.slice(2), {
+  boolean: ['verbose', 'commit'],
+  string: ['batchSize'],
+  default: {
+    batchSize: '1000',
+  },
+})
+
+const batchSize = parseInt(args.batchSize, 10)
+
+/**
+ * @typedef {import('ioredis').Redis} Redis
+ */
+
+/**
+ *
+ * @param {string} key
+ * @return {string|void}
+ */
+function extractDocId(key) {
+  const matches = key.match(/ProjectHistoryId:\{(.*?)\}/)
+  if (matches) {
+    return matches[1]
+  }
+}
+
+/**
+ *
+ * @param {string} docId
+ * @return {Promise<{projectId: string, historyId: string}>}
+ */
+async function getHistoryId(docId) {
+  const doc = await db.docs.findOne(
+    { _id: new ObjectId(docId) },
+    { projection: { project_id: 1 }, readPreference: 'secondaryPreferred' }
+  )
+
+  if (!doc) {
+    throw new OError('Doc not present in mongo', { docId })
+  }
+
+  const project = await db.projects.findOne(
+    { _id: doc.project_id },
+    {
+      projection: { 'overleaf.history': 1 },
+      readPreference: 'secondaryPreferred',
+    }
+  )
+
+  if (!project?.overleaf?.history?.id) {
+    throw new OError('Project not present in mongo (or has no history id)', {
+      docId,
+      project,
+      doc,
+    })
+  }
+
+  return {
+    historyId: project?.overleaf?.history?.id,
+    projectId: doc.project_id.toString(),
+  }
+}
+
+/**
+ * @typedef {Object} UpdateableDoc
+ * @property {string} docId
+ * @property {string} projectId
+ * @property {string} historyId
+ */
+
+/**
+ *
+ * @param {Redis} node
+ * @param {Array<string>} docIds
+ * @return {Promise<Array<UpdateableDoc>>}
+ */
+async function findDocsWithMissingHistoryIds(node, docIds) {
+  const historyIds = await node.mget(
+    docIds.map(docId => docUpdaterKeys.projectHistoryId({ doc_id: docId }))
+  )
+
+  const results = []
+
+  for (const index in docIds) {
+    const historyId = historyIds[index]
+    const docId = docIds[index]
+    if (!historyId) {
+      try {
+        const { projectId, historyId } = await getHistoryId(docId)
+        results.push({ projectId, historyId, docId })
+      } catch (error) {
+        logger.warn(
+          { error },
+          'Error gathering data for doc with missing history id'
+        )
+      }
+    }
+  }
+  return results
+}
+
+/**
+ *
+ * @param {Array<UpdateableDoc>} updates
+ * @return {Promise<void>}
+ */
+async function fixAndFlushProjects(updates) {
+  for (const update of updates) {
+    logger.debug({ ...update }, 'Flushing project')
+    if (commit) {
+      await rclient.set(
+        docUpdaterKeys.projectHistoryId({ doc_id: '*' }),
+        update.historyId
+      )
+      await ProjectManager.promises.flushAndDeleteProjectWithLocks(
+        update.projectId,
+        {}
+      )
+    }
+  }
+}
+
+/**
+ *
+ * @param {Array<Redis>} nodes
+ * @param {number} batchSize
+ * @return {Promise<void>}
+ */
+async function scanNodes(nodes, batchSize = 1000) {
+  let scanned = 0
+
+  for (const node of nodes) {
+    const stream = node.scanStream({
+      match: docUpdaterKeys.projectHistoryId({ doc_id: '*' }),
+      count: batchSize,
+    })
+
+    for await (const docKeys of stream) {
+      if (docKeys.length === 0) {
+        continue
+      }
+      stream.pause()
+      scanned += docKeys.length
+
+      const docIds = docKeys
+        .map((/** @type {string} */ docKey) => extractDocId(docKey))
+        .filter(Boolean)
+
+      try {
+        const updates = await findDocsWithMissingHistoryIds(node, docIds)
+        if (updates.length > 0) {
+          logger.info({ updates }, 'Found doc(s) with missing history ids')
+          await fixAndFlushProjects(updates)
+        }
+      } catch (error) {
+        logger.error({ docKeys }, 'Error processing batch')
+      } finally {
+        stream.resume()
+      }
+    }
+
+    logger.info({ scanned, server: node.serverInfo.role }, 'Scanned node')
+  }
+}
+
+async function main({ batchSize }) {
+  const nodes = (typeof rclient.nodes === 'function'
+    ? rclient.nodes('master')
+    : undefined) || [rclient]
+  await scanNodes(nodes, batchSize)
+}
+
+let code = 0
+
+main({ batchSize })
+  .then(() => {
+    logger.info({}, 'done')
+  })
+  .catch(error => {
+    logger.error({ error }, 'error')
+    code = 1
+  })
+  .finally(() => {
+    rclient.quit().then(() => process.exit(code))
+  })