Details
-
Bug
-
Resolution: Fixed
-
Critical
-
4.0.0
-
Security Level: Public
-
400-4019
System test cluster : 10M and 100 warehouses.
-
Untriaged
-
Yes
-
Mar 9 - Mar 27
Description
1. Last 2 phases on the System testing.
- Reboot entire cluster.
Couple of observations
- indexes on bucket default ( active and replica indexes) seem to be rebuilt.
example: index fare_idx,
items remaining on index pre reboot:0
post reboot: 10M
likewise for every index for bucket default.
- Additionally, the index active resident ratio drops from 50-60% to 20-25%
Unlike KV where the access logs/ last accessed infromation is used to maximize active KV resident ratio.
attached screenshot of the cluster before/after reboot
attached logs from node 233, index node for default
attaching logs from all nodes.
Attachments
For Gerrit Dashboard: MB-15983 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
54305,5 | Merge remote-tracking branch 'couchbase/sherlock' | master | ep-engine | Status: MERGED | +2 | +1 |
54318,4 | MB-15983: During warmup, add lastSnapEnd in vb failover logs if possible. | sherlock | ep-engine | Status: MERGED | +2 | +1 |