Details
-
Bug
-
Resolution: Cannot Reproduce
-
Major
-
5.5.0
-
32GB RAM
Samsung PM863a (SATA SSD)
-
Untriaged
-
Centos 64-bit
-
-
Unknown
Description
Build 5.5.0-1682.
Setup:
- 2 nodes
- 10GB RAM quota
- 1 bucket, full eviction
The ingestion process got stuck after inserting 116M items.
Replication from *.100 to *.101 doesn't seem to be working correctly. It falls behind and eventually gets blocked once resident ratio on *.101 drops to 0. *.100 receives nothing but backoff while memory usage on *.101 remains constantly above the high watermark.