overleaf/services/document-updater/app/coffee/RedisManager.coffee

319 lines
14 KiB
CoffeeScript
Raw Normal View History

2014-02-12 05:40:42 -05:00
Settings = require('settings-sharelatex')
rclient = require("redis-sharelatex").createClient(Settings.redis.documentupdater)
2014-02-12 05:40:42 -05:00
_ = require('underscore')
logger = require('logger-sharelatex')
metrics = require('./Metrics')
Errors = require "./Errors"
2017-02-14 11:11:43 -05:00
crypto = require "crypto"
2018-03-09 09:14:14 -05:00
ProjectHistoryRedisManager = require "./ProjectHistoryRedisManager"
# Sometimes Redis calls take an unexpectedly long time. We have to be
# quick with Redis calls because we're holding a lock that expires
# after 30 seconds. We can't let any errors in the rest of the stack
# hold us up, and need to bail out quickly if there is a problem.
MAX_REDIS_REQUEST_LENGTH = 5000 # 5 seconds
# Make times easy to read
minutes = 60 # seconds for Redis expire
2017-02-17 07:29:15 -05:00
# LUA script to write document and return hash
# arguments: docLinesKey docLines
setScript = """
redis.call('set', KEYS[1], ARGV[1])
return redis.sha1hex(ARGV[1])
"""
2017-02-22 08:46:34 -05:00
logHashErrors = Settings.documentupdater?.logHashErrors
2017-02-22 09:32:51 -05:00
logHashReadErrors = logHashErrors?.read
logHashWriteErrors = logHashErrors?.write
2017-02-22 08:46:34 -05:00
MEGABYTES = 1024 * 1024
MAX_RANGES_SIZE = 3 * MEGABYTES
keys = Settings.redis.documentupdater.key_schema
2017-05-08 10:56:02 -04:00
historyKeys = Settings.redis.history.key_schema
module.exports = RedisManager =
rclient: rclient
putDocInMemory : (project_id, doc_id, docLines, version, ranges, pathname, projectHistoryId, _callback)->
2014-02-12 05:40:42 -05:00
timer = new metrics.Timer("redis.put-doc")
callback = (error) ->
2014-02-12 05:40:42 -05:00
timer.done()
_callback(error)
2017-02-14 11:11:43 -05:00
docLines = JSON.stringify(docLines)
if docLines.indexOf("\u0000") != -1
error = new Error("null bytes found in doc lines")
logger.error {err: error, doc_id: doc_id, docLines: docLines}, error.message
return callback(error)
2017-02-14 11:11:43 -05:00
docHash = RedisManager._computeHash(docLines)
logger.log {project_id, doc_id, version, docHash, pathname, projectHistoryId}, "putting doc in redis"
RedisManager._serializeRanges ranges, (error, ranges) ->
if error?
logger.error {err: error, doc_id, project_id}, error.message
return callback(error)
multi = rclient.multi()
multi.eval setScript, 1, keys.docLines(doc_id:doc_id), docLines
multi.set keys.projectKey({doc_id:doc_id}), project_id
multi.set keys.docVersion(doc_id:doc_id), version
multi.set keys.docHash(doc_id:doc_id), docHash
if ranges?
multi.set keys.ranges(doc_id:doc_id), ranges
else
multi.del keys.ranges(doc_id:doc_id)
2017-09-29 05:54:48 -04:00
multi.set keys.pathname(doc_id:doc_id), pathname
multi.set keys.projectHistoryId(doc_id:doc_id), projectHistoryId
multi.exec (error, result) ->
return callback(error) if error?
# check the hash computed on the redis server
writeHash = result?[0]
if logHashWriteErrors and writeHash? and writeHash isnt docHash
logger.error project_id: project_id, doc_id: doc_id, writeHash: writeHash, origHash: docHash, docLines:docLines, "hash mismatch on putDocInMemory"
# update docsInProject set
rclient.sadd keys.docsInProject(project_id:project_id), doc_id, callback
2014-02-12 05:40:42 -05:00
removeDocFromMemory : (project_id, doc_id, _callback)->
2014-02-12 05:40:42 -05:00
logger.log project_id:project_id, doc_id:doc_id, "removing doc from redis"
callback = (err) ->
2014-02-12 05:40:42 -05:00
if err?
logger.err project_id:project_id, doc_id:doc_id, err:err, "error removing doc from redis"
_callback(err)
2014-02-12 05:40:42 -05:00
else
logger.log project_id:project_id, doc_id:doc_id, "removed doc from redis"
_callback()
multi = rclient.multi()
multi.del keys.docLines(doc_id:doc_id)
multi.del keys.projectKey(doc_id:doc_id)
multi.del keys.docVersion(doc_id:doc_id)
2017-02-14 11:11:43 -05:00
multi.del keys.docHash(doc_id:doc_id)
multi.del keys.ranges(doc_id:doc_id)
2017-09-29 05:54:48 -04:00
multi.del keys.pathname(doc_id:doc_id)
multi.del keys.projectHistoryId(doc_id:doc_id)
2017-10-06 07:23:23 -04:00
multi.del keys.unflushedTime(doc_id:doc_id)
multi.exec (error) ->
return callback(error) if error?
multi = rclient.multi()
multi.srem keys.docsInProject(project_id:project_id), doc_id
multi.del keys.projectState(project_id:project_id)
multi.exec callback
2014-02-12 05:40:42 -05:00
checkOrSetProjectState: (project_id, newState, callback = (error, stateChanged) ->) ->
multi = rclient.multi()
multi.getset keys.projectState(project_id:project_id), newState
multi.expire keys.projectState(project_id:project_id), 30 * minutes
multi.exec (error, response) ->
return callback(error) if error?
logger.log project_id: project_id, newState:newState, oldState: response[0], "checking project state"
callback(null, response[0] isnt newState)
2017-09-08 10:50:26 -04:00
clearProjectState: (project_id, callback = (error) ->) ->
rclient.del keys.projectState(project_id:project_id), callback
getDoc : (project_id, doc_id, callback = (error, lines, version, ranges, pathname, projectHistoryId, unflushedTime) ->)->
2014-02-12 05:40:42 -05:00
timer = new metrics.Timer("redis.get-doc")
2016-06-01 05:53:50 -04:00
multi = rclient.multi()
multi.get keys.docLines(doc_id:doc_id)
2014-02-12 05:40:42 -05:00
multi.get keys.docVersion(doc_id:doc_id)
2017-02-14 11:11:43 -05:00
multi.get keys.docHash(doc_id:doc_id)
2016-09-02 09:47:41 -04:00
multi.get keys.projectKey(doc_id:doc_id)
multi.get keys.ranges(doc_id:doc_id)
2017-09-29 05:54:48 -04:00
multi.get keys.pathname(doc_id:doc_id)
multi.get keys.projectHistoryId(doc_id:doc_id)
2017-10-06 07:23:23 -04:00
multi.get keys.unflushedTime(doc_id:doc_id)
multi.exec (error, [docLines, version, storedHash, doc_project_id, ranges, pathname, projectHistoryId, unflushedTime])->
timeSpan = timer.done()
2014-02-12 05:40:42 -05:00
return callback(error) if error?
# check if request took too long and bail out. only do this for
# get, because it is the first call in each update, so if this
# passes we'll assume others have a reasonable chance to succeed.
2017-06-26 11:36:05 -04:00
if timeSpan > MAX_REDIS_REQUEST_LENGTH
error = new Error("redis getDoc exceeded timeout")
return callback(error)
2017-02-15 09:12:36 -05:00
# check sha1 hash value if present
if docLines? and storedHash?
2017-02-14 11:11:43 -05:00
computedHash = RedisManager._computeHash(docLines)
2017-02-22 09:32:51 -05:00
if logHashReadErrors and computedHash isnt storedHash
2017-03-03 10:57:44 -05:00
logger.error project_id: project_id, doc_id: doc_id, doc_project_id: doc_project_id, computedHash: computedHash, storedHash: storedHash, docLines:docLines, "hash mismatch on retrieved document"
2017-02-14 11:11:43 -05:00
2016-06-01 05:53:50 -04:00
try
docLines = JSON.parse docLines
ranges = RedisManager._deserializeRanges(ranges)
2016-06-01 05:53:50 -04:00
catch e
return callback(e)
version = parseInt(version or 0, 10)
2016-09-02 09:47:41 -04:00
# check doc is in requested project
if doc_project_id? and doc_project_id isnt project_id
logger.error project_id: project_id, doc_id: doc_id, doc_project_id: doc_project_id, "doc not in project"
return callback(new Errors.NotFoundError("document not found"))
if projectHistoryId?
projectHistoryId = parseInt(projectHistoryId)
# doc is not in redis, bail out
2017-05-31 11:08:33 -04:00
if !docLines?
return callback null, docLines, version, ranges, pathname, projectHistoryId, unflushedTime
# doc should be in project set, check if missing (workaround for missing docs from putDoc)
rclient.sadd keys.docsInProject(project_id:project_id), doc_id, (error, result) ->
return callback(error) if error?
if result isnt 0 # doc should already be in set
logger.error project_id: project_id, doc_id: doc_id, doc_project_id: doc_project_id, "doc missing from docsInProject set"
callback null, docLines, version, ranges, pathname, projectHistoryId, unflushedTime
2014-02-12 05:40:42 -05:00
getDocVersion: (doc_id, callback = (error, version) ->) ->
rclient.get keys.docVersion(doc_id: doc_id), (error, version) ->
return callback(error) if error?
version = parseInt(version, 10)
callback null, version
getDocLines: (doc_id, callback = (error, version) ->) ->
rclient.get keys.docLines(doc_id: doc_id), (error, docLines) ->
return callback(error) if error?
callback null, docLines
2014-02-12 05:40:42 -05:00
getPreviousDocOps: (doc_id, start, end, callback = (error, jsonOps) ->) ->
timer = new metrics.Timer("redis.get-prev-docops")
2014-02-12 05:40:42 -05:00
rclient.llen keys.docOps(doc_id: doc_id), (error, length) ->
return callback(error) if error?
rclient.get keys.docVersion(doc_id: doc_id), (error, version) ->
return callback(error) if error?
version = parseInt(version, 10)
first_version_in_redis = version - length
if start < first_version_in_redis or end > version
error = new Errors.OpRangeNotAvailableError("doc ops range is not loaded in redis")
logger.warn {err: error, doc_id, length, version, start, end}, "doc ops range is not loaded in redis"
2014-02-12 05:40:42 -05:00
return callback(error)
start = start - first_version_in_redis
if end > -1
end = end - first_version_in_redis
if isNaN(start) or isNaN(end)
error = new Error("inconsistent version or lengths")
logger.error {err: error, doc_id, length, version, start, end}, "inconsistent version or length"
2014-02-12 05:40:42 -05:00
return callback(error)
rclient.lrange keys.docOps(doc_id: doc_id), start, end, (error, jsonOps) ->
return callback(error) if error?
try
ops = jsonOps.map (jsonOp) -> JSON.parse jsonOp
catch e
return callback(e)
timeSpan = timer.done()
2017-06-26 11:36:05 -04:00
if timeSpan > MAX_REDIS_REQUEST_LENGTH
error = new Error("redis getPreviousDocOps exceeded timeout")
return callback(error)
2014-02-12 05:40:42 -05:00
callback null, ops
DOC_OPS_TTL: 60 * minutes
DOC_OPS_MAX_LENGTH: 100
updateDocument : (project_id, doc_id, docLines, newVersion, appliedOps = [], ranges, callback = (error) ->)->
RedisManager.getDocVersion doc_id, (error, currentVersion) ->
2014-02-12 05:40:42 -05:00
return callback(error) if error?
if currentVersion + appliedOps.length != newVersion
error = new Error("Version mismatch. '#{doc_id}' is corrupted.")
logger.error {err: error, doc_id, currentVersion, newVersion, opsLength: appliedOps.length}, "version mismatch"
return callback(error)
jsonOps = appliedOps.map (op) -> JSON.stringify op
2017-05-22 10:20:28 -04:00
for op in jsonOps
if op.indexOf("\u0000") != -1
error = new Error("null bytes found in jsonOps")
logger.error {err: error, doc_id: doc_id, jsonOps: jsonOps}, error.message
2017-05-22 10:20:28 -04:00
return callback(error)
2017-05-22 09:58:38 -04:00
2017-02-14 11:11:43 -05:00
newDocLines = JSON.stringify(docLines)
if newDocLines.indexOf("\u0000") != -1
error = new Error("null bytes found in doc lines")
logger.error {err: error, doc_id: doc_id, newDocLines: newDocLines}, error.message
return callback(error)
2017-02-14 11:11:43 -05:00
newHash = RedisManager._computeHash(newDocLines)
2017-05-08 10:56:02 -04:00
opVersions = appliedOps.map (op) -> op?.v
logger.info doc_id: doc_id, version: newVersion, hash: newHash, op_versions: opVersions, "updating doc in redis"
2017-05-08 10:56:02 -04:00
RedisManager._serializeRanges ranges, (error, ranges) ->
if error?
logger.error {err: error, doc_id}, error.message
return callback(error)
2017-05-22 09:58:51 -04:00
if ranges? and ranges.indexOf("\u0000") != -1
error = new Error("null bytes found in ranges")
logger.error err: error, doc_id: doc_id, ranges: ranges, error.message
return callback(error)
multi = rclient.multi()
multi.eval setScript, 1, keys.docLines(doc_id:doc_id), newDocLines # index 0
multi.set keys.docVersion(doc_id:doc_id), newVersion # index 1
multi.set keys.docHash(doc_id:doc_id), newHash # index 2
multi.ltrim keys.docOps(doc_id: doc_id), -RedisManager.DOC_OPS_MAX_LENGTH, -1 # index 3
if ranges?
multi.set keys.ranges(doc_id:doc_id), ranges # index 4
else
multi.del keys.ranges(doc_id:doc_id) # also index 4
# push the ops last so we can get the lengths at fixed index position 7
if jsonOps.length > 0
multi.rpush keys.docOps(doc_id: doc_id), jsonOps... # index 5
# expire must come after rpush since before it will be a no-op if the list is empty
multi.expire keys.docOps(doc_id: doc_id), RedisManager.DOC_OPS_TTL # index 6
multi.rpush historyKeys.uncompressedHistoryOps(doc_id: doc_id), jsonOps... # index 7
2017-10-06 07:23:23 -04:00
# Set the unflushed timestamp to the current time if the doc
# hasn't been modified before (the content in mongo has been
# valid up to this point). Otherwise leave it alone ("NX" flag).
multi.set keys.unflushedTime(doc_id: doc_id), Date.now(), "NX"
multi.exec (error, result) ->
return callback(error) if error?
# check the hash computed on the redis server
writeHash = result?[0]
if logHashWriteErrors and writeHash? and writeHash isnt newHash
logger.error doc_id: doc_id, writeHash: writeHash, origHash: newHash, docLines:newDocLines, "hash mismatch on updateDocument"
# length of uncompressedHistoryOps queue (index 7)
docUpdateCount = result[7]
if jsonOps.length > 0 && Settings.apis?.project_history?.enabled
2018-03-09 09:14:14 -05:00
ProjectHistoryRedisManager.queueOps project_id, jsonOps..., (error, projectUpdateCount) ->
callback null, docUpdateCount, projectUpdateCount
else
callback null, docUpdateCount
renameDoc: (project_id, doc_id, user_id, update, projectHistoryId, callback = (error) ->) ->
2017-11-06 11:14:27 -05:00
RedisManager.getDoc project_id, doc_id, (error, lines, version) ->
return callback(error) if error?
if lines? and version?
rclient.set keys.pathname(doc_id:doc_id), update.newPathname, (error) ->
return callback(error) if error?
ProjectHistoryRedisManager.queueRenameEntity project_id, projectHistoryId, 'doc', doc_id, user_id, update, callback
2017-11-06 11:14:27 -05:00
else
ProjectHistoryRedisManager.queueRenameEntity project_id, projectHistoryId, 'doc', doc_id, user_id, update, callback
2014-02-12 05:40:42 -05:00
2017-10-06 07:23:23 -04:00
clearUnflushedTime: (doc_id, callback = (error) ->) ->
rclient.del keys.unflushedTime(doc_id:doc_id), callback
2014-02-12 05:40:42 -05:00
getDocIdsInProject: (project_id, callback = (error, doc_ids) ->) ->
rclient.smembers keys.docsInProject(project_id: project_id), callback
_serializeRanges: (ranges, callback = (error, serializedRanges) ->) ->
jsonRanges = JSON.stringify(ranges)
if jsonRanges? and jsonRanges.length > MAX_RANGES_SIZE
return callback new Error("ranges are too large")
if jsonRanges == '{}'
# Most doc will have empty ranges so don't fill redis with lots of '{}' keys
jsonRanges = null
return callback null, jsonRanges
_deserializeRanges: (ranges) ->
if !ranges? or ranges == ""
return {}
else
2017-02-14 11:11:43 -05:00
return JSON.parse(ranges)
_computeHash: (docLines) ->
2017-02-21 11:03:06 -05:00
# use sha1 checksum of doclines to detect data corruption.
#
# note: must specify 'utf8' encoding explicitly, as the default is
# binary in node < v5
return crypto.createHash('sha1').update(docLines, 'utf8').digest('hex')