Merge pull request #4352 from overleaf/sk-add-ce-sp-migration-deleted-files

Add migrations for CE/SP, run scripts for deleted files and docs

GitOrigin-RevId: e425a8d929a4aefdc6f61cff636fb5a1b4c67dce

services/web/migrations/20210727123346_ce_sp_backfill_deleted_files.js

@@ -0,0 +1,14 @@
+const runScript = require('../scripts/back_fill_deleted_files.js')
+
+exports.tags = ['server-ce', 'server-pro', 'saas']
+
+exports.migrate = async client => {
+  const options = {
+    performCleanup: true,
+    letUserDoubleCheckInputsFor: 10,
+    fixPartialInserts: true,
+  }
+  await runScript(options)
+}
+
+exports.rollback = async client => {}

services/web/migrations/20210727150530_ce_sp_backfill_deleted_docs.js

@@ -0,0 +1,13 @@
+const runScript = require('../scripts/back_fill_doc_name_for_deleted_docs.js')
+
+exports.tags = ['server-ce', 'server-pro', 'saas']
+
+exports.migrate = async client => {
+  const options = {
+    performCleanup: true,
+    letUserDoubleCheckInputsFor: 10,
+  }
+  await runScript(options)
+}
+
+exports.rollback = async client => {}

services/web/migrations/20210728115327_ce_sp_backfill_dummy_doc_meta.js

@@ -0,0 +1,23 @@
+const runScript = require('../scripts/back_fill_dummy_doc_meta.js')
+
+exports.tags = ['server-ce', 'server-pro', 'saas']
+
+exports.migrate = async client => {
+  const { db } = client
+  const [firstProject] = await db.projects
+    .find()
+    .sort({ _id: 1 })
+    .limit(1)
+    .toArray()
+  if (!firstProject) {
+    return
+  }
+  const options = {
+    firstProjectId: firstProject._id,
+    performCleanup: true,
+    letUserDoubleCheckInputsFor: 10,
+  }
+  await runScript(options)
+}
+
+exports.rollback = async client => {}

services/web/migrations/lib/adapter.js

@@ -7,7 +7,11 @@ const {
 
 class Adapter {
   constructor(params) {
-    if (!(process.argv.includes('-t') || process.argv.includes('--tags'))) {
+    if (
+      !process.env.SKIP_TAG_CHECK &&
+      !process.argv.includes('create') &&
+      !(process.argv.includes('-t') || process.argv.includes('--tags'))
+    ) {
       console.error("ERROR: must pass tags using '-t' or '--tags', exiting")
       process.exit(1)
     }

services/web/scripts/back_fill_deleted_files.js

@@ -1,58 +1,62 @@
-const WRITE_CONCURRENCY = parseInt(process.env.WRITE_CONCURRENCY, 10) || 10
-
 const { batchedUpdate } = require('./helpers/batchedUpdate')
 const { promiseMapWithLimit, promisify } = require('../app/src/util/promises')
 const { db } = require('../app/src/infrastructure/mongodb')
 const sleep = promisify(setTimeout)
+const _ = require('lodash')
 
-const PERFORM_CLEANUP = process.argv.includes('--perform-cleanup')
-const FIX_PARTIAL_INSERTS = process.argv.includes('--fix-partial-inserts')
-const LET_USER_DOUBLE_CHECK_INPUTS_FOR = parseInt(
+async function main(options) {
+  if (!options) {
+    options = {}
+  }
+  _.defaults(options, {
+    writeConcurrency: parseInt(process.env.WRITE_CONCURRENCY, 10) || 10,
+    performCleanup: process.argv.includes('--perform-cleanup'),
+    fixPartialInserts: process.argv.includes('--fix-partial-inserts'),
+    letUserDoubleCheckInputsFor: parseInt(
       process.env.LET_USER_DOUBLE_CHECK_INPUTS_FOR || 10 * 1000,
       10
-)
+    ),
+  })
 
-async function main() {
-  await letUserDoubleCheckInputs()
+  await letUserDoubleCheckInputs(options)
 
   await batchedUpdate(
     'projects',
     // array is not empty ~ array has one item
     { 'deletedFiles.0': { $exists: true } },
-    processBatch,
+    async (x, projects) => {
+      await processBatch(x, projects, options)
+    },
     { _id: 1, deletedFiles: 1 }
   )
 }
 
-main()
-  .then(() => {
-    process.exit(0)
-  })
-  .catch(error => {
-    console.error({ error })
-    process.exit(1)
-  })
-
-async function processBatch(_, projects) {
-  await promiseMapWithLimit(WRITE_CONCURRENCY, projects, processProject)
+async function processBatch(_, projects, options) {
+  await promiseMapWithLimit(
+    options.writeConcurrency,
+    projects,
+    async project => {
+      await processProject(project, options)
+    }
+  )
 }
 
-async function processProject(project) {
-  await backFillFiles(project)
+async function processProject(project, options) {
+  await backFillFiles(project, options)
 
-  if (PERFORM_CLEANUP) {
+  if (options.performCleanup) {
     await cleanupProject(project)
   }
 }
 
-async function backFillFiles(project) {
+async function backFillFiles(project, options) {
   const projectId = project._id
   filterDuplicatesInPlace(project)
   project.deletedFiles.forEach(file => {
     file.projectId = projectId
   })
 
-  if (FIX_PARTIAL_INSERTS) {
+  if (options.fixPartialInserts) {
     await fixPartialInserts(project)
   } else {
     await db.deletedFiles.insertMany(project.deletedFiles)
@@ -98,8 +102,8 @@ async function cleanupProject(project) {
   )
 }
 
-async function letUserDoubleCheckInputs() {
-  if (PERFORM_CLEANUP) {
+async function letUserDoubleCheckInputs(options) {
+  if (options.performCleanup) {
     console.error('BACK FILLING AND PERFORMING CLEANUP')
   } else {
     console.error(
@@ -108,8 +112,21 @@ async function letUserDoubleCheckInputs() {
   }
   console.error(
     'Waiting for you to double check inputs for',
-    LET_USER_DOUBLE_CHECK_INPUTS_FOR,
+    options.letUserDoubleCheckInputsFor,
     'ms'
   )
-  await sleep(LET_USER_DOUBLE_CHECK_INPUTS_FOR)
+  await sleep(options.letUserDoubleCheckInputsFor)
+}
+
+module.exports = main
+
+if (require.main === module) {
+  main()
+    .then(() => {
+      process.exit(0)
+    })
+    .catch(error => {
+      console.error({ error })
+      process.exit(1)
+    })
 }

services/web/scripts/back_fill_doc_name_for_deleted_docs.js

@@ -1,46 +1,50 @@
-const WRITE_CONCURRENCY = parseInt(process.env.WRITE_CONCURRENCY, 10) || 10
-
 const { batchedUpdate } = require('./helpers/batchedUpdate')
 const { promiseMapWithLimit, promisify } = require('../app/src/util/promises')
 const { db } = require('../app/src/infrastructure/mongodb')
 const sleep = promisify(setTimeout)
+const _ = require('lodash')
 
-const PERFORM_CLEANUP = process.argv.pop() === '--perform-cleanup'
-const LET_USER_DOUBLE_CHECK_INPUTS_FOR = parseInt(
+async function main(options) {
+  if (!options) {
+    options = {}
+  }
+  _.defaults(options, {
+    writeConcurrency: parseInt(process.env.WRITE_CONCURRENCY, 10) || 10,
+    performCleanup: process.argv.pop() === '--perform-cleanup',
+    letUserDoubleCheckInputsFor: parseInt(
       process.env.LET_USER_DOUBLE_CHECK_INPUTS_FOR || 10 * 1000,
       10
-)
+    ),
+  })
 
-async function main() {
-  await letUserDoubleCheckInputs()
+  await letUserDoubleCheckInputs(options)
 
   await batchedUpdate(
     'projects',
     // array is not empty ~ array has one item
     { 'deletedDocs.0': { $exists: true } },
-    processBatch,
+    async (_collection, projects) => {
+      await processBatch(_collection, projects, options)
+    },
     { _id: 1, deletedDocs: 1 }
   )
 }
 
-main()
-  .then(() => {
-    process.exit(0)
-  })
-  .catch(error => {
-    console.error({ error })
-    process.exit(1)
-  })
-
-async function processBatch(_, projects) {
-  await promiseMapWithLimit(WRITE_CONCURRENCY, projects, processProject)
+async function processBatch(_, projects, options) {
+  await promiseMapWithLimit(
+    options.writeConcurrency,
+    projects,
+    async project => {
+      await processProject(project, options)
+    }
+  )
 }
 
-async function processProject(project) {
+async function processProject(project, options) {
   for (const doc of project.deletedDocs) {
     await backFillDoc(doc)
   }
-  if (PERFORM_CLEANUP) {
+  if (options.performCleanup) {
     await cleanupProject(project)
   }
 }
@@ -57,8 +61,8 @@ async function cleanupProject(project) {
   )
 }
 
-async function letUserDoubleCheckInputs() {
-  if (PERFORM_CLEANUP) {
+async function letUserDoubleCheckInputs(options) {
+  if (options.performCleanup) {
     console.error('BACK FILLING AND PERFORMING CLEANUP')
   } else {
     console.error(
@@ -67,8 +71,21 @@ async function letUserDoubleCheckInputs() {
   }
   console.error(
     'Waiting for you to double check inputs for',
-    LET_USER_DOUBLE_CHECK_INPUTS_FOR,
+    options.letUserDoubleCheckInputsFor,
     'ms'
   )
-  await sleep(LET_USER_DOUBLE_CHECK_INPUTS_FOR)
+  await sleep(options.letUserDoubleCheckInputsFor)
+}
+
+module.exports = main
+
+if (require.main === module) {
+  main()
+    .then(() => {
+      process.exit(0)
+    })
+    .catch(error => {
+      console.error({ error })
+      process.exit(1)
+    })
 }

services/web/scripts/back_fill_dummy_doc_meta.js

@@ -2,45 +2,54 @@ const { promisify } = require('util')
 const { ObjectId, ReadPreference } = require('mongodb')
 const { db, waitForDb } = require('../app/src/infrastructure/mongodb')
 const sleep = promisify(setTimeout)
+const _ = require('lodash')
 
 const NOW_IN_S = Date.now() / 1000
 const ONE_WEEK_IN_S = 60 * 60 * 24 * 7
 const TEN_SECONDS = 10 * 1000
 
-const CACHE_SIZE = parseInt(process.env.CACHE_SIZE, 10) || 100
-const DRY_RUN = process.env.DRY_RUN === 'true'
-if (!process.env.FIRST_PROJECT_ID) {
-  console.error('Set FIRST_PROJECT_ID and re-run.')
-  process.exit(1)
-}
-const FIRST_PROJECT_ID = ObjectId(process.env.FIRST_PROJECT_ID)
-const INCREMENT_BY_S = parseInt(process.env.INCREMENT_BY_S, 10) || ONE_WEEK_IN_S
-const BATCH_SIZE = parseInt(process.env.BATCH_SIZE, 10) || 1000
-const STOP_AT_S = parseInt(process.env.STOP_AT_S, 10) || NOW_IN_S
-const LET_USER_DOUBLE_CHECK_INPUTS_FOR =
-  parseInt(process.env.LET_USER_DOUBLE_CHECK_INPUTS_FOR, 10) || TEN_SECONDS
-
 const DUMMY_NAME = 'unknown.tex'
 const DUMMY_TIME = new Date('2021-04-12T00:00:00.000Z')
 
 const LRUCache = require('lru-cache')
-const deletedProjectsCache = new LRUCache({
-  max: CACHE_SIZE,
-})
+let deletedProjectsCache = null
 
 function getSecondsFromObjectId(id) {
   return id.getTimestamp().getTime() / 1000
 }
 
-async function main() {
-  await letUserDoubleCheckInputs()
+async function main(options) {
+  if (!options) {
+    options = {}
+  }
+  _.defaults(options, {
+    dryRun: process.env.DRY_RUN === 'true',
+    cacheSize: parseInt(process.env.CACHE_SIZE, 10) || 100,
+    firstProjectId: ObjectId(process.env.FIRST_PROJECT_ID),
+    incrementByS: parseInt(process.env.INCREMENT_BY_S, 10) || ONE_WEEK_IN_S,
+    batchSize: parseInt(process.env.BATCH_SIZE, 10) || 1000,
+    stopAtS: parseInt(process.env.STOP_AT_S, 10) || NOW_IN_S,
+    letUserDoubleCheckInputsFor:
+      parseInt(process.env.LET_USER_DOUBLE_CHECK_INPUTS_FOR, 10) || TEN_SECONDS,
+  })
+
+  if (!options.firstProjectId) {
+    console.error('Set FIRST_PROJECT_ID and re-run.')
+    process.exit(1)
+  }
+
+  deletedProjectsCache = new LRUCache({
+    max: options.cacheSize,
+  })
+
+  await letUserDoubleCheckInputs(options)
   await waitForDb()
 
-  let startId = FIRST_PROJECT_ID
+  let startId = options.firstProjectId
 
   let nProcessed = 0
-  while (getSecondsFromObjectId(startId) <= STOP_AT_S) {
-    const end = getSecondsFromObjectId(startId) + INCREMENT_BY_S
+  while (getSecondsFromObjectId(startId) <= options.stopAtS) {
+    const end = getSecondsFromObjectId(startId) + options.incrementByS
     let endId = ObjectId.createFromTime(end)
     const query = {
       project_id: {
@@ -57,16 +66,16 @@ async function main() {
     const docs = await db.docs
       .find(query, { readPreference: ReadPreference.SECONDARY })
       .project({ _id: 1, project_id: 1 })
-      .limit(BATCH_SIZE)
+      .limit(options.batchSize)
       .toArray()
 
     if (docs.length) {
       const docIds = docs.map(doc => doc._id)
       console.log('Back filling dummy meta data for', JSON.stringify(docIds))
-      await processBatch(docs)
+      await processBatch(docs, options)
       nProcessed += docIds.length
 
-      if (docs.length === BATCH_SIZE) {
+      if (docs.length === options.batchSize) {
         endId = docs[docs.length - 1].project_id
       }
     }
@@ -94,7 +103,7 @@ async function getDeletedProject(projectId) {
   return deletedProject
 }
 
-async function processBatch(docs) {
+async function processBatch(docs, options) {
   for (const doc of docs) {
     const { _id: docId, project_id: projectId } = doc
     const deletedProject = await getDeletedProject(projectId)
@@ -119,37 +128,25 @@ async function processBatch(docs) {
     } else {
       console.log('Orphaned deleted doc %s (no deletedProjects entry)', docId)
     }
-    if (DRY_RUN) continue
+    if (options.dryRun) continue
     await db.docs.updateOne({ _id: docId }, { $set: { name, deletedAt } })
   }
 }
 
-async function letUserDoubleCheckInputs() {
-  console.error(
-    'Options:',
-    JSON.stringify(
-      {
-        BATCH_SIZE,
-        CACHE_SIZE,
-        DRY_RUN,
-        FIRST_PROJECT_ID,
-        INCREMENT_BY_S,
-        STOP_AT_S,
-        LET_USER_DOUBLE_CHECK_INPUTS_FOR,
-      },
-      null,
-      2
-    )
-  )
+async function letUserDoubleCheckInputs(options) {
+  console.error('Options:', JSON.stringify(options, null, 2))
   console.error(
     'Waiting for you to double check inputs for',
-    LET_USER_DOUBLE_CHECK_INPUTS_FOR,
+    options.letUserDoubleCheckInputsFor,
     'ms'
   )
-  await sleep(LET_USER_DOUBLE_CHECK_INPUTS_FOR)
+  await sleep(options.letUserDoubleCheckInputsFor)
 }
 
-main()
+module.exports = main
+
+if (require.main === module) {
+  main()
     .then(() => {
       console.error('Done.')
       process.exit(0)
@@ -158,3 +155,4 @@ main()
       console.error({ error })
       process.exit(1)
     })
+}

services/web/scripts/mark_migration.js

@@ -0,0 +1,67 @@
+const Adapter = require('../migrations/lib/adapter')
+const fs = require('fs').promises
+const path = require('path')
+
+async function main(args) {
+  if (
+    !args ||
+    args.length === 0 ||
+    args.includes('help') ||
+    args.includes('--help') ||
+    args.includes('-h')
+  ) {
+    console.log('')
+    console.log('usage: node ./scripts/mark_migration.js migration state')
+    console.log('')
+    console.log('    migration:   name of migration file')
+    console.log('    state:       executed | unexecuted')
+    console.log('')
+    return
+  }
+
+  const migration = args[0]
+  if (!migration) {
+    throw new Error('Error: migration must be supplied')
+  }
+  const state = args[1]
+  if (!state) {
+    throw new Error('Error: migration state must be supplied')
+  }
+
+  try {
+    await fs.access(path.join(__dirname, '../migrations', `${migration}.js`))
+  } catch (err) {
+    throw new Error(
+      `Error: migration ${migration} does not exist on disk: ${err}`
+    )
+  }
+
+  console.log(`Marking ${migration} as ${state}`)
+
+  process.env.SKIP_TAG_CHECK = 'true'
+  const adapter = new Adapter()
+  await adapter.connect()
+  switch (state) {
+    case 'executed':
+      await adapter.markExecuted(migration)
+      break
+    case 'unexecuted':
+      await adapter.unmarkExecuted(migration)
+      break
+    default:
+      throw new Error(`invalid state "${state}"`)
+  }
+  console.log('Done')
+}
+
+if (require.main === module) {
+  const args = process.argv.slice(2)
+  main(args)
+    .then(() => {
+      process.exit(0)
+    })
+    .catch(err => {
+      console.error(err)
+      process.exit(1)
+    })
+}

services/web/test/acceptance/src/BackFillDummyDocMetaTests.js

@@ -188,7 +188,15 @@ describe('BackFillDummyDocMeta', function () {
       '',
     ])
     expect(stdErr.filter(filterOutput)).to.deep.equal([
-      ...`Options: ${JSON.stringify(options, null, 2)}`.split('\n'),
+      `Options: {`,
+      `  "dryRun": ${options.DRY_RUN},`,
+      `  "cacheSize": ${options.CACHE_SIZE},`,
+      `  "firstProjectId": "${options.FIRST_PROJECT_ID}",`,
+      `  "incrementByS": ${options.INCREMENT_BY_S},`,
+      `  "batchSize": ${options.BATCH_SIZE},`,
+      `  "stopAtS": ${options.STOP_AT_S},`,
+      `  "letUserDoubleCheckInputsFor": ${options.LET_USER_DOUBLE_CHECK_INPUTS_FOR}`,
+      '}',
       'Waiting for you to double check inputs for 1 ms',
       `Processed 1 until ${getObjectIdFromDate('2021-04-02T00:00:00.000Z')}`,
       `Processed 2 until ${getObjectIdFromDate('2021-04-03T00:00:00.000Z')}`,