3_pack_docHistory_collection.coffee 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341
  1. Settings = require "settings-sharelatex"
  2. fs = require("fs")
  3. mongojs = require("mongojs")
  4. ObjectId = mongojs.ObjectId
  5. db = mongojs(Settings.mongo.url, ['docs','docHistory', 'docHistoryStats'])
  6. _ = require("underscore")
  7. async = require("async")
  8. exec = require("child_process").exec
  9. BSON = db.bson.BSON
  10. logger = {
  11. log: ->
  12. err: ->
  13. }
  14. needToExit = false
  15. handleExit = () ->
  16. needToExit = true
  17. console.log('Got signal. Shutting down.')
  18. process.on 'SIGINT', handleExit
  19. process.on 'SIGHUP', handleExit
  20. finished_docs_path = "/tmp/finished-docs-3"
  21. all_docs_path = "/tmp/all-docs-3"
  22. unmigrated_docs_path = "/tmp/unmigrated-docs-3"
  23. finished_docs = {}
  24. if fs.existsSync(finished_docs_path)
  25. for id in fs.readFileSync(finished_docs_path,'utf-8').split("\n")
  26. finished_docs[id] = true
  27. getAndWriteDocids = (callback)->
  28. console.log "finding all doc id's - #{new Date().toString()}"
  29. db.docs.find {}, {_id:1}, (err, ids)->
  30. console.log "total found docs in mongo #{ids.length} - #{new Date().toString()}"
  31. ids = _.pluck ids, '_id'
  32. ids = _.filter ids, (id)-> id?
  33. fileData = ids.join("\n")
  34. fs.writeFileSync all_docs_path + ".tmp", fileData
  35. fs.renameSync all_docs_path + ".tmp", all_docs_path
  36. callback(err, ids)
  37. loadDocIds = (callback)->
  38. console.log "loading doc ids from #{all_docs_path}"
  39. data = fs.readFileSync all_docs_path, "utf-8"
  40. ids = data.split("\n")
  41. console.log "loaded #{ids.length} doc ids from #{all_docs_path}"
  42. callback null, ids
  43. getDocIds = (callback)->
  44. exists = fs.existsSync all_docs_path
  45. if exists
  46. loadDocIds callback
  47. else
  48. getAndWriteDocids callback
  49. markDocAsProcessed = (doc_id, callback)->
  50. finished_docs[doc_id] = true
  51. fs.appendFile finished_docs_path, "#{doc_id}\n", callback
  52. markDocAsUnmigrated = (doc_id, callback)->
  53. console.log "#{doc_id} unmigrated"
  54. markDocAsProcessed doc_id, (err)->
  55. fs.appendFile unmigrated_docs_path, "#{doc_id}\n", callback
  56. checkIfDocHasBeenProccessed = (doc_id, callback)->
  57. callback(null, finished_docs[doc_id])
  58. processNext = (doc_id, callback)->
  59. if !doc_id? or doc_id.length == 0
  60. return callback()
  61. if needToExit
  62. return callback(new Error("graceful shutdown"))
  63. checkIfDocHasBeenProccessed doc_id, (err, hasBeenProcessed)->
  64. if hasBeenProcessed
  65. console.log "#{doc_id} already processed, skipping"
  66. return callback()
  67. PackManager._packDocHistory doc_id, {}, (err) ->
  68. if err?
  69. console.log "error processing #{doc_id}"
  70. markDocAsUnmigrated doc_id, callback
  71. else
  72. markDocAsProcessed doc_id, callback
  73. updateIndexes = (callback) ->
  74. async.series [
  75. (cb) ->
  76. console.log "create index"
  77. db.docHistory.ensureIndex { project_id: 1, "meta.end_ts": 1, "meta.start_ts": -1 }, { background: true }, cb
  78. (cb) ->
  79. console.log "drop index"
  80. db.docHistory.dropIndex { project_id: 1, "meta.end_ts": 1 }, cb
  81. (cb) ->
  82. console.log "drop index"
  83. db.docHistory.dropIndex { project_id: 1, "pack.0.meta.end_ts": 1, "meta.end_ts": 1}, cb
  84. ], (err, results) ->
  85. console.log "all done"
  86. callback(err)
  87. exports.migrate = (client, done = ->)->
  88. getDocIds (err, ids)->
  89. totalDocCount = ids.length
  90. alreadyFinishedCount = Object.keys(finished_docs).length
  91. t0 = Date.now()
  92. printProgress = () ->
  93. count = Object.keys(finished_docs).length
  94. processedFraction = (count-alreadyFinishedCount)/totalDocCount
  95. remainingFraction = (totalDocCount-count)/totalDocCount
  96. t = Date.now()
  97. dt = (t-t0)*remainingFraction/processedFraction
  98. estFinishTime = new Date(t + dt)
  99. console.log "completed #{count}/#{totalDocCount} processed=#{processedFraction.toFixed(2)} remaining=#{remainingFraction.toFixed(2)} elapsed=#{(t-t0)/1000} est Finish=#{estFinishTime}"
  100. interval = setInterval printProgress, 3*1000
  101. nextId = null
  102. testFn = () ->
  103. return false if needToExit
  104. id = ids.shift()
  105. while id? and finished_docs[id] # skip finished
  106. id = ids.shift()
  107. nextId = id
  108. return nextId?
  109. executeFn = (cb) ->
  110. processNext nextId, cb
  111. async.whilst testFn, executeFn, (err)->
  112. if err?
  113. console.error err, "at end of jobs"
  114. else
  115. console.log "finished at #{new Date}"
  116. clearInterval interval
  117. done(err)
  118. exports.rollback = (client, done)->
  119. done()
  120. # process.nextTick () ->
  121. # exports.migrate () ->
  122. # console.log "done"
  123. DAYS = 24 * 3600 * 1000 # one day in milliseconds
  124. # copied from track-changes/app/coffee/PackManager.coffee
  125. PackManager =
  126. MAX_SIZE: 1024*1024 # make these configurable parameters
  127. MAX_COUNT: 512
  128. convertDocsToPacks: (docs, callback) ->
  129. packs = []
  130. top = null
  131. docs.forEach (d,i) ->
  132. # skip existing packs
  133. if d.pack?
  134. top = null
  135. return
  136. sz = BSON.calculateObjectSize(d)
  137. # decide if this doc can be added to the current pack
  138. validLength = top? && (top.pack.length < PackManager.MAX_COUNT)
  139. validSize = top? && (top.sz + sz < PackManager.MAX_SIZE)
  140. bothPermanent = top? && (top.expiresAt? is false) && (d.expiresAt? is false)
  141. bothTemporary = top? && (top.expiresAt? is true) && (d.expiresAt? is true)
  142. within1Day = bothTemporary && (d.meta.start_ts - top.meta.start_ts < 24 * 3600 * 1000)
  143. if top? && validLength && validSize && (bothPermanent || (bothTemporary && within1Day))
  144. top.pack = top.pack.concat {v: d.v, meta: d.meta, op: d.op, _id: d._id}
  145. top.sz += sz
  146. top.n += 1
  147. top.v_end = d.v
  148. top.meta.end_ts = d.meta.end_ts
  149. top.expiresAt = d.expiresAt if top.expiresAt?
  150. return
  151. else
  152. # create a new pack
  153. top = _.clone(d)
  154. top.pack = [ {v: d.v, meta: d.meta, op: d.op, _id: d._id} ]
  155. top.meta = { start_ts: d.meta.start_ts, end_ts: d.meta.end_ts }
  156. top.sz = sz
  157. top.n = 1
  158. top.v_end = d.v
  159. delete top.op
  160. delete top._id
  161. packs.push top
  162. callback(null, packs)
  163. checkHistory: (docs, callback) ->
  164. errors = []
  165. prev = null
  166. error = (args...) ->
  167. errors.push args
  168. docs.forEach (d,i) ->
  169. if d.pack?
  170. n = d.pack.length
  171. last = d.pack[n-1]
  172. error('bad pack v_end', d) if d.v_end != last.v
  173. error('bad pack start_ts', d) if d.meta.start_ts != d.pack[0].meta.start_ts
  174. error('bad pack end_ts', d) if d.meta.end_ts != last.meta.end_ts
  175. d.pack.forEach (p, i) ->
  176. prev = v
  177. v = p.v
  178. error('bad version', v, 'in', p) if v <= prev
  179. #error('expired op', p, 'in pack') if p.expiresAt?
  180. else
  181. prev = v
  182. v = d.v
  183. error('bad version', v, 'in', d) if v <= prev
  184. if errors.length
  185. callback(errors)
  186. else
  187. callback()
  188. insertPack: (packObj, callback) ->
  189. bulk = db.docHistory.initializeOrderedBulkOp()
  190. doc_id = packObj.doc_id
  191. expect_nInserted = 1
  192. expect_nRemoved = packObj.pack.length
  193. logger.log {doc_id: doc_id}, "adding pack, removing #{expect_nRemoved} ops"
  194. bulk.insert packObj
  195. ids = (op._id for op in packObj.pack)
  196. bulk.find({_id:{$in:ids}}).remove()
  197. bulk.execute (err, result) ->
  198. if err?
  199. logger.error {doc_id: doc_id}, "error adding pack"
  200. callback(err, result)
  201. else if result.nInserted != expect_nInserted or result.nRemoved != expect_nRemoved
  202. logger.error {doc_id: doc_id, result}, "unexpected result adding pack"
  203. callback(new Error(
  204. msg: 'unexpected result'
  205. expected: {expect_nInserted, expect_nRemoved}
  206. ), result)
  207. else
  208. db.docHistoryStats.update {doc_id:doc_id}, {
  209. $inc:{update_count:-expect_nRemoved},
  210. $currentDate:{last_packed:true}
  211. }, {upsert:true}, () ->
  212. callback(err, result)
  213. # retrieve document ops/packs and check them
  214. getDocHistory: (doc_id, callback) ->
  215. db.docHistory.find({doc_id:ObjectId(doc_id)}).sort {v:1}, (err, docs) ->
  216. return callback(err) if err?
  217. # for safety, do a consistency check of the history
  218. logger.log {doc_id}, "checking history for document"
  219. PackManager.checkHistory docs, (err) ->
  220. return callback(err) if err?
  221. callback(err, docs)
  222. #PackManager.deleteExpiredPackOps docs, (err) ->
  223. # return callback(err) if err?
  224. # callback err, docs
  225. packDocHistory: (doc_id, options, callback) ->
  226. if typeof callback == "undefined" and typeof options == 'function'
  227. callback = options
  228. options = {}
  229. LockManager.runWithLock(
  230. "HistoryLock:#{doc_id}",
  231. (releaseLock) ->
  232. PackManager._packDocHistory(doc_id, options, releaseLock)
  233. , callback
  234. )
  235. _packDocHistory: (doc_id, options, callback) ->
  236. logger.log {doc_id},"starting pack operation for document history"
  237. PackManager.getDocHistory doc_id, (err, docs) ->
  238. return callback(err) if err?
  239. origDocs = 0
  240. origPacks = 0
  241. for d in docs
  242. if d.pack? then origPacks++ else origDocs++
  243. PackManager.convertDocsToPacks docs, (err, packs) ->
  244. return callback(err) if err?
  245. total = 0
  246. for p in packs
  247. total = total + p.pack.length
  248. logger.log {doc_id, origDocs, origPacks, newPacks: packs.length, totalOps: total}, "document stats"
  249. if packs.length
  250. if options['dry-run']
  251. logger.log {doc_id}, 'dry-run, skipping write packs'
  252. return callback()
  253. PackManager.savePacks packs, (err) ->
  254. return callback(err) if err?
  255. # check the history again
  256. PackManager.getDocHistory doc_id, callback
  257. else
  258. logger.log {doc_id}, "no packs to write"
  259. # keep a record that we checked this one to avoid rechecking it
  260. db.docHistoryStats.update {doc_id:doc_id}, {
  261. $currentDate:{last_checked:true}
  262. }, {upsert:true}, () ->
  263. callback null, null
  264. DB_WRITE_DELAY: 100
  265. savePacks: (packs, callback) ->
  266. async.eachSeries packs, PackManager.safeInsert, (err, result) ->
  267. if err?
  268. logger.log {err, result}, "error writing packs"
  269. callback err, result
  270. else
  271. callback()
  272. safeInsert: (packObj, callback) ->
  273. PackManager.insertPack packObj, (err, result) ->
  274. setTimeout () ->
  275. callback(err,result)
  276. , PackManager.DB_WRITE_DELAY
  277. deleteExpiredPackOps: (docs, callback) ->
  278. now = Date.now()
  279. toRemove = []
  280. toUpdate = []
  281. docs.forEach (d,i) ->
  282. if d.pack?
  283. newPack = d.pack.filter (op) ->
  284. if op.expiresAt? then op.expiresAt > now else true
  285. if newPack.length == 0
  286. toRemove.push d
  287. else if newPack.length < d.pack.length
  288. # adjust the pack properties
  289. d.pack = newPack
  290. first = d.pack[0]
  291. last = d.pack[d.pack.length - 1]
  292. d.v_end = last.v
  293. d.meta.start_ts = first.meta.start_ts
  294. d.meta.end_ts = last.meta.end_ts
  295. toUpdate.push d
  296. if toRemove.length or toUpdate.length
  297. bulk = db.docHistory.initializeOrderedBulkOp()
  298. toRemove.forEach (pack) ->
  299. console.log "would remove", pack
  300. #bulk.find({_id:pack._id}).removeOne()
  301. toUpdate.forEach (pack) ->
  302. console.log "would update", pack
  303. #bulk.find({_id:pack._id}).updateOne(pack);
  304. bulk.execute callback
  305. else
  306. callback()