-
Type: Bug
-
Resolution: Duplicate
-
Priority: Major - P3
-
None
-
Affects Version/s: None
-
Component/s: Replication
-
None
-
ALL
The primary of a replica set of my shard was killed due to an OOM issue.
After that the server is not able to restart.
Version is 3.2.3 WiredTiger
2017-01-13T16:54:54.263+0000 I NETWORK [initandlisten] connection accepted from 54.78.134.204:57528 #12 (11 connections now open) 2017-01-13T16:54:54.294+0000 I ASIO [NetworkInterfaceASIO-BGSync-0] Successfully connected to:27017 2017-01-13T16:54:54.607+0000 I ACCESS [conn12] Successfully authenticated as principal __system on local 2017-01-13T16:54:54.673+0000 I REPL [rsBackgroundSync] Starting rollback due to OplogStartMissing: our last op time fetched: (term: 405, timestamp: Jan 13 16:09:01:3). source's GTE: (term: 406, ti mestamp: Jan 13 16:09:01:3) hashes: (1329492769709081641/-2280498797367626345) 2017-01-13T16:54:54.673+0000 I - [rsBackgroundSync] Fatal assertion 18750 UnrecoverableRollbackError: need to rollback, but in inconsistent state. minvalid: (term: 406, timestamp: Jan 13 16:13: 31:f) > our last optime: (term: 405, timestamp: Jan 13 16:09:01:3) 2017-01-13T16:54:54.673+0000 I - [rsBackgroundSync] ***aborting after fassert() failure
- duplicates
-
SERVER-25145 During rollback (or w/minvalid invalid) select sync source based on end OpTime
- Closed
- related to
-
SERVER-27149 Sync source selection doesn't consider terms
- Closed