Shane Kilkelly
|
c3465cf3d4
|
Reduce retry delay to 10ms, and refactor
|
2017-07-18 15:28:18 +01:00 |
|
Shane Kilkelly
|
0d46c3e2b3
|
WIP: auto-retry web requests
|
2017-07-17 10:08:21 +01:00 |
|
Brian Gough
|
c21cc013f3
|
use null instead of err in callbacks on success
|
2017-07-12 10:48:32 +01:00 |
|
Brian Gough
|
bb0dc4b4d5
|
release the lock if it took too long to acquire it
|
2017-07-12 10:48:32 +01:00 |
|
Brian Gough
|
54c0fc1180
|
put a 5 second timeout in getting redis lock
|
2017-07-11 14:28:58 +01:00 |
|
Brian Gough
|
ffea31290c
|
return total time from the profile .end() method
|
2017-07-11 14:28:58 +01:00 |
|
Brian Gough
|
59ace9d5c5
|
avoid long lines in timeout check
|
2017-06-26 16:36:05 +01:00 |
|
Brian Gough
|
963e513057
|
add a timeout for getPreviousDocOps
it uses several redis operations and this makes it prone to subsequent
timeouts if getDoc succeeds but is slow
|
2017-06-23 15:50:21 +01:00 |
|
James Allen
|
54f128289f
|
Merge pull request #50 from sharelatex/bg-metric-for-unlock-errors
add metric for unlock errors
|
2017-06-13 14:46:40 +01:00 |
|
James Allen
|
7d22e4422c
|
Merge pull request #49 from sharelatex/bg-add-redis-timeout
add redis timeout
|
2017-06-13 14:46:32 +01:00 |
|
James Allen
|
a3d7260619
|
Ensure expires is called after the key is created
|
2017-06-08 16:43:06 +01:00 |
|
Brian Gough
|
2e5d57cd99
|
don't increase rate limit when tasks are failing
|
2017-06-07 11:47:28 +01:00 |
|
Brian Gough
|
1ef258c878
|
add a timeout on the redis getDoc request
|
2017-06-07 11:34:42 +01:00 |
|
Brian Gough
|
673f4228ce
|
add metric for unlock errors
sentry does not record them reliably, due to rate limiting
|
2017-06-05 16:29:58 +01:00 |
|
Brian Gough
|
12e8eaa9b6
|
fix bug in doclines check
|
2017-05-31 16:08:33 +01:00 |
|
Brian Gough
|
bed64d26c8
|
check if doc is missing from DocsIn set
|
2017-05-31 15:33:59 +01:00 |
|
Brian Gough
|
fa089dc28d
|
avoid graphite drift in active worker count
|
2017-05-30 14:37:01 +01:00 |
|
Brian Gough
|
891ffda3bf
|
put a rate limit on worker calls
|
2017-05-26 15:08:35 +01:00 |
|
Brian Gough
|
c43f97e8e3
|
add metric for active workers
|
2017-05-24 12:12:15 +01:00 |
|
Brian Gough
|
a3947c587f
|
fix bug, jsonOps is array not string
|
2017-05-22 15:20:28 +01:00 |
|
Brian Gough
|
54ad45c870
|
add null byte check for ranges
|
2017-05-22 15:00:05 +01:00 |
|
Brian Gough
|
a13f055d62
|
add null byte check for ops
|
2017-05-22 15:00:00 +01:00 |
|
Brian Gough
|
78828ebbc5
|
fine grained logging for cpu usage in applyUpdate
|
2017-05-19 16:00:16 +01:00 |
|
Brian Gough
|
a550078071
|
replace verbose logging by slow query log
|
2017-05-18 15:04:12 +01:00 |
|
Brian Gough
|
95596061d4
|
clarify calculation of process.hrtime
|
2017-05-18 15:02:54 +01:00 |
|
Brian Gough
|
e00f4dde8e
|
log profilers start and end times for reference
|
2017-05-18 15:02:44 +01:00 |
|
Brian Gough
|
e11f64f83a
|
increase profiling cutoff to 1s
|
2017-05-18 15:02:08 +01:00 |
|
Brian Gough
|
b6efb051e1
|
add profiling
|
2017-05-18 11:17:08 +01:00 |
|
Brian Gough
|
387ecac6cb
|
log for each get/try/release lock
|
2017-05-18 11:17:08 +01:00 |
|
Brian Gough
|
dff6e2c3da
|
bug fix in backoff
|
2017-05-16 16:31:28 +01:00 |
|
Brian Gough
|
c7d8fbbb8a
|
exponential backoff for lock up to maximum time
|
2017-05-16 15:22:02 +01:00 |
|
Brian Gough
|
b15d2ef796
|
Revert "fail safely if doc cannot be loaded"
|
2017-05-15 16:18:40 +01:00 |
|
Paulo Jorge Reis
|
2c3f0aa6b4
|
Merge pull request #34 from sharelatex/pr-bulk-actions
Add methods to bulk accept changes.
|
2017-05-15 15:14:33 +01:00 |
|
Paulo Reis
|
ef7fa59257
|
Update ranges tracker.
|
2017-05-15 11:11:14 +01:00 |
|
Paulo Reis
|
05d7d1b8c6
|
Merge branch 'master' into pr-bulk-actions
|
2017-05-15 10:35:16 +01:00 |
|
Brian Gough
|
2ee40d0748
|
ensure document is in redis before consuming ops
|
2017-05-12 14:22:18 +01:00 |
|
Paulo Reis
|
5677c7ad9a
|
Update range tracker.
|
2017-05-11 11:31:19 +01:00 |
|
Brian Gough
|
36407ac726
|
rename HistoryManager pushUncompressedHistoryOps
|
2017-05-09 12:02:27 +01:00 |
|
Brian Gough
|
7ce6285e3d
|
increase flush threshold to 100 ops
|
2017-05-09 10:44:26 +01:00 |
|
Brian Gough
|
fdf5e8e0b8
|
get history ops length directly from redis update
|
2017-05-09 10:34:31 +01:00 |
|
Brian Gough
|
2d158b03d7
|
rename pushUncompressedHistoryOps
|
2017-05-09 09:32:56 +01:00 |
|
Paulo Reis
|
144804e516
|
Update tests with new methods.
|
2017-05-08 16:34:17 +01:00 |
|
Paulo Reis
|
d7ac57d054
|
Use new bulk method.
|
2017-05-08 16:08:42 +01:00 |
|
Paulo Reis
|
70341e4e80
|
Add bulk method from range tracker.
|
2017-05-08 16:08:34 +01:00 |
|
Brian Gough
|
79d8fced49
|
make history update more atomic
|
2017-05-08 15:58:29 +01:00 |
|
Paulo Reis
|
e96e6d6a29
|
Remove unused method.
|
2017-05-05 15:17:19 +01:00 |
|
Paulo Reis
|
9b8a2c3030
|
Consolidate methods to accept changes.
|
2017-05-05 15:12:06 +01:00 |
|
Paulo Reis
|
d530ee9501
|
Add methods to bulk accept changes.
|
2017-05-04 15:32:54 +01:00 |
|
James Allen
|
4104ca4889
|
Add in separate redis config for the lock and fix a few web -> realtime
|
2017-05-03 16:27:32 +01:00 |
|
James Allen
|
ed493d8ad3
|
Rename web -> realtime for consistency with realtime
|
2017-05-02 15:38:33 +01:00 |
|