ProjectPersistenceManager.js 7.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242
  1. // TODO: This file was created by bulk-decaffeinate.
  2. // Fix any style issues and re-enable lint.
  3. /*
  4. * decaffeinate suggestions:
  5. * DS101: Remove unnecessary use of Array.from
  6. * DS102: Remove unnecessary code created because of implicit returns
  7. * DS207: Consider shorter variations of null checks
  8. * Full docs: https://github.com/decaffeinate/decaffeinate/blob/master/docs/suggestions.md
  9. */
  10. let ProjectPersistenceManager
  11. const UrlCache = require('./UrlCache')
  12. const CompileManager = require('./CompileManager')
  13. const async = require('async')
  14. const logger = require('@overleaf/logger')
  15. const oneDay = 24 * 60 * 60 * 1000
  16. const Metrics = require('@overleaf/metrics')
  17. const Settings = require('@overleaf/settings')
  18. const diskusage = require('diskusage')
  19. const { callbackify } = require('node:util')
  20. const Path = require('node:path')
  21. const fs = require('node:fs')
  22. // projectId -> timestamp mapping.
  23. const LAST_ACCESS = new Map()
  24. async function collectDiskStats() {
  25. const paths = [
  26. Settings.path.compilesDir,
  27. Settings.path.outputDir,
  28. Settings.path.clsiCacheDir,
  29. ]
  30. const diskStats = {}
  31. for (const path of paths) {
  32. try {
  33. const stats = await diskusage.check(path)
  34. const diskAvailablePercent = (stats.available / stats.total) * 100
  35. Metrics.gauge('disk_available_percent', diskAvailablePercent, 1, {
  36. path,
  37. })
  38. const lowDisk = diskAvailablePercent < 10
  39. diskStats[path] = { stats, lowDisk }
  40. } catch (err) {
  41. logger.err({ err, path }, 'error getting disk usage')
  42. }
  43. }
  44. return diskStats
  45. }
  46. async function refreshExpiryTimeout() {
  47. for (const [path, { stats, lowDisk }] of Object.entries(
  48. await collectDiskStats()
  49. )) {
  50. const lowerExpiry = ProjectPersistenceManager.EXPIRY_TIMEOUT * 0.9
  51. if (lowDisk && Settings.project_cache_length_ms / 2 < lowerExpiry) {
  52. logger.warn(
  53. {
  54. path,
  55. stats,
  56. newExpiryTimeoutInDays: (lowerExpiry / oneDay).toFixed(2),
  57. },
  58. 'disk running low on space, modifying EXPIRY_TIMEOUT'
  59. )
  60. ProjectPersistenceManager.EXPIRY_TIMEOUT = lowerExpiry
  61. break
  62. }
  63. }
  64. }
  65. module.exports = ProjectPersistenceManager = {
  66. EXPIRY_TIMEOUT: Settings.project_cache_length_ms || oneDay * 2.5,
  67. promises: {
  68. refreshExpiryTimeout,
  69. },
  70. refreshExpiryTimeout: callbackify(refreshExpiryTimeout),
  71. init() {
  72. fs.readdir(Settings.path.compilesDir, (err, dirs) => {
  73. if (err) {
  74. logger.warn({ err }, 'cannot get project listing')
  75. dirs = []
  76. }
  77. async.eachLimit(
  78. dirs,
  79. 10,
  80. (projectAndUserId, cb) => {
  81. const compileDir = Path.join(
  82. Settings.path.compilesDir,
  83. projectAndUserId
  84. )
  85. const projectId = projectAndUserId.slice(0, 24)
  86. fs.stat(compileDir, (err, stats) => {
  87. if (err) {
  88. // Schedule for immediate cleanup
  89. LAST_ACCESS.set(projectId, 0)
  90. } else {
  91. // Cleanup eventually.
  92. LAST_ACCESS.set(projectId, stats.mtime.getTime())
  93. }
  94. cb()
  95. })
  96. },
  97. () => {
  98. setInterval(
  99. () => {
  100. ProjectPersistenceManager.refreshExpiryTimeout(() => {
  101. ProjectPersistenceManager.clearExpiredProjects(err => {
  102. if (err) {
  103. logger.error({ err }, 'clearing expired projects failed')
  104. }
  105. })
  106. })
  107. },
  108. 10 * 60 * 1000
  109. )
  110. }
  111. )
  112. })
  113. // Collect disk stats frequently to have them ready the next time /metrics is scraped (60s +- jitter).
  114. setInterval(() => {
  115. collectDiskStats().catch(err => {
  116. logger.err({ err }, 'low level error collecting disk stats')
  117. })
  118. }, 50_000)
  119. },
  120. markProjectAsJustAccessed(projectId, callback) {
  121. LAST_ACCESS.set(projectId, Date.now())
  122. callback()
  123. },
  124. clearExpiredProjects(callback) {
  125. if (callback == null) {
  126. callback = function () {}
  127. }
  128. return ProjectPersistenceManager._findExpiredProjectIds(
  129. function (error, projectIds) {
  130. if (error != null) {
  131. return callback(error)
  132. }
  133. logger.debug({ projectIds }, 'clearing expired projects')
  134. const jobs = Array.from(projectIds || []).map(projectId =>
  135. (
  136. projectId => callback =>
  137. ProjectPersistenceManager.clearProjectFromCache(
  138. projectId,
  139. { reason: 'expired' },
  140. function (err) {
  141. if (err != null) {
  142. logger.error({ err, projectId }, 'error clearing project')
  143. }
  144. return callback()
  145. }
  146. )
  147. )(projectId)
  148. )
  149. return async.series(jobs, function (error) {
  150. if (error != null) {
  151. return callback(error)
  152. }
  153. return CompileManager.clearExpiredProjects(
  154. ProjectPersistenceManager.EXPIRY_TIMEOUT,
  155. error => callback(error)
  156. )
  157. })
  158. }
  159. )
  160. }, // ignore any errors from deleting directories
  161. clearProject(projectId, userId, callback) {
  162. if (callback == null) {
  163. callback = function () {}
  164. }
  165. logger.debug({ projectId, userId }, 'clearing project for user')
  166. return CompileManager.clearProject(projectId, userId, function (error) {
  167. if (error != null) {
  168. return callback(error)
  169. }
  170. return ProjectPersistenceManager.clearProjectFromCache(
  171. projectId,
  172. { reason: 'cleared' },
  173. function (error) {
  174. if (error != null) {
  175. return callback(error)
  176. }
  177. return callback()
  178. }
  179. )
  180. })
  181. },
  182. clearProjectFromCache(projectId, options, callback) {
  183. if (callback == null) {
  184. callback = function () {}
  185. }
  186. logger.debug({ projectId }, 'clearing project from cache')
  187. return UrlCache.clearProject(projectId, options, function (error) {
  188. if (error != null) {
  189. logger.err({ error, projectId }, 'error clearing project from cache')
  190. return callback(error)
  191. }
  192. return ProjectPersistenceManager._clearProjectFromDatabase(
  193. projectId,
  194. function (error) {
  195. if (error != null) {
  196. logger.err(
  197. { error, projectId },
  198. 'error clearing project from database'
  199. )
  200. }
  201. return callback(error)
  202. }
  203. )
  204. })
  205. },
  206. _clearProjectFromDatabase(projectId, callback) {
  207. LAST_ACCESS.delete(projectId)
  208. callback()
  209. },
  210. _findExpiredProjectIds(callback) {
  211. const expiredFrom = Date.now() - ProjectPersistenceManager.EXPIRY_TIMEOUT
  212. const expiredProjectsIds = []
  213. for (const [projectId, lastAccess] of LAST_ACCESS.entries()) {
  214. if (lastAccess < expiredFrom) {
  215. expiredProjectsIds.push(projectId)
  216. }
  217. }
  218. // ^ may be a fairly busy loop, continue detached.
  219. setTimeout(() => callback(null, expiredProjectsIds), 0)
  220. },
  221. }
  222. logger.debug(
  223. { EXPIRY_TIMEOUT: ProjectPersistenceManager.EXPIRY_TIMEOUT },
  224. 'project assets kept timeout'
  225. )