| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134 |
- const Settings = require('@overleaf/settings')
- const { fetchJson } = require('@overleaf/fetch-utils')
- const { waitForDb } = require('../app/src/infrastructure/mongodb')
- const { promiseMapWithLimit } = require('@overleaf/promise-utils')
- const { getHardDeletedProjectIds } = require('./delete_orphaned_data_helper')
- const TpdsUpdateSender = require('../app/src/Features/ThirdPartyDataStore/TpdsUpdateSender')
- const { promisify } = require('util')
- const { ObjectId } = require('mongodb')
- const sleep = promisify(setTimeout)
- const START_OFFSET = process.env.START_OFFSET
- const BATCH_SIZE = parseInt(process.env.BATCH_SIZE, 10) || 1000
- const DRY_RUN = process.env.DRY_RUN !== 'false'
- const READ_CONCURRENCY_SECONDARY =
- parseInt(process.env.READ_CONCURRENCY_SECONDARY, 10) || 1000
- const READ_CONCURRENCY_PRIMARY =
- parseInt(process.env.READ_CONCURRENCY_PRIMARY, 10) || 500
- const WRITE_CONCURRENCY = parseInt(process.env.WRITE_CONCURRENCY, 10) || 10
- const LET_USER_DOUBLE_CHECK_INPUTS_FOR =
- parseInt(process.env.LET_USER_DOUBLE_CHECK_INPUTS_FOR, 10) || 10 * 1000
- async function main() {
- await letUserDoubleCheckInputs()
- await waitForDb()
- let processed = 0
- let hardDeleted = 0
- let pageToken = ''
- let startOffset = START_OFFSET
- while (pageToken !== undefined) {
- const url = new URL(`${Settings.apis.project_archiver.url}/project/list`)
- url.searchParams.append('pageToken', pageToken)
- url.searchParams.append('startOffset', startOffset)
- const { nextPageToken, entries } = await fetchJson(url)
- pageToken = nextPageToken
- startOffset = undefined
- hardDeleted += await processBatch(entries)
- processed += entries.length
- console.log(
- 'processed:',
- processed.toString().padStart(10, '0'),
- 'hard deleted:',
- hardDeleted.toString().padStart(10, '0'),
- 'nextPageToken:',
- nextPageToken,
- 'START_OFFSET:',
- entries.pop()?.prefix
- )
- }
- }
- async function processBatch(entries) {
- const projectIdToPrefix = new Map()
- for (const { prefix, projectId } of entries) {
- const prefixes = projectIdToPrefix.get(projectId) || []
- prefixes.push(prefix)
- projectIdToPrefix.set(projectId, prefixes)
- }
- const projectIds = Array.from(projectIdToPrefix.keys()).map(
- id => new ObjectId(id)
- )
- const projectsWithOrphanedArchive = await getHardDeletedProjectIds({
- projectIds,
- READ_CONCURRENCY_PRIMARY,
- READ_CONCURRENCY_SECONDARY,
- })
- await promiseMapWithLimit(
- WRITE_CONCURRENCY,
- projectsWithOrphanedArchive.flatMap(id =>
- projectIdToPrefix.get(id.toString())
- ),
- hardDeleteProjectArchiverData
- )
- return projectsWithOrphanedArchive.length
- }
- async function hardDeleteProjectArchiverData(prefix) {
- console.log(`Destroying hard deleted project archive at '${prefix}/'`)
- if (DRY_RUN) return
- for (let i = 0; i < 10; i++) {
- await sleep(1000 * i)
- try {
- const ok = await TpdsUpdateSender.promises.deleteProject({
- projectId: encodeURIComponent(prefix),
- })
- if (ok) {
- return
- }
- } catch (e) {
- console.error(`deletion failed for '${prefix}/'`, e)
- }
- }
- throw new Error(`deletion failed for '${prefix}/', check logs`)
- }
- async function letUserDoubleCheckInputs() {
- console.error(
- 'Options:',
- JSON.stringify(
- {
- BATCH_SIZE,
- DRY_RUN,
- LET_USER_DOUBLE_CHECK_INPUTS_FOR,
- READ_CONCURRENCY_SECONDARY,
- READ_CONCURRENCY_PRIMARY,
- START_OFFSET,
- WRITE_CONCURRENCY,
- },
- null,
- 2
- )
- )
- console.error(
- 'Waiting for you to double check inputs for',
- LET_USER_DOUBLE_CHECK_INPUTS_FOR,
- 'ms'
- )
- await sleep(LET_USER_DOUBLE_CHECK_INPUTS_FOR)
- }
- main()
- .then(() => {
- console.log('Done.')
- process.exit(0)
- })
- .catch(error => {
- console.error({ error })
- process.exit(1)
- })
|