Details
-
Bug
-
Resolution: Fixed
-
Critical
-
Cheshire-Cat
-
Untriaged
-
-
1
-
Unknown
Description
Build : 7.0.0-4432
Test : -test tests/integration/cheshirecat/test_cheshirecat_kv_gsi_coll_xdcr_backup_sgw_fts_itemct_txns_eventing_cbas.yml -scope tests/integration/cheshirecat/scope_cheshirecat_with_backup.yml
Scale : 3
Iteration : 1st
The N1QL workload to run queries against collection started at 2021-02-10T16:48:07. There are 2 N1QL nodes in the cluster - 172.23.97.149 & 172.23.97.150.
At 21:05, 172.23.97.150 got OOM killed by the OS and 21:14, 172.23.97.149 got OOM killed.
The attached graphs for 172.23.97.149 & 172.23.97.150 shows that the available memory gradually declined from the time the N1QL workload was started and ultimately got OOM killed. After restarting, we can see the gradual decline start again. The N1QL request rate is mostly constant along this time.
Attached logs for review.
Attachments
Issue Links
- is duplicated by
-
MB-44376 cbq-engine triggering Linux OOM-killer during rebalance
- Closed