Details
-
Bug
-
Resolution: Duplicate
-
Major
-
3.0
-
Security Level: Public
-
None
-
centos 6.4 64-bit
-
Untriaged
-
Centos 64-bit
-
-
Yes
Description
Install couchbase server 3.0.0-858 on 8 nodes
1:172.23.107.195
2:172.23.107.196
3:172.23.107.197
4:172.23.107.199
5:172.23.107.200
6:172.23.107.202
7:172.23.107.201
8:172.23.107.203
Create a cluster of 7 nodes
Create 2 buckets: default and sasl-2 (no view)
Load 25 M items to each bucket to bring down active resident ratio down to 80%
Do update, expired and delete on both buckets in 3 hours.
Then add node 203 to cluster. Passed
Remove node 200 out cluster. Passed
Swap master node (remove 195 and add back 200). Passed
Failover node 201. Rebalance hang.
Check bucket sasl-2, see mem used > high water mark
Adjust bucket quota on bucket sasl-2 to make mem used < high water mark.
Rebalance still hang
Check memory used. Total mem used of cluster < high water mark, but at each node, some nodes have mem used higher than water mark.
This bug may relate to bug MB-11488
Attachments
Issue Links
- relates to
-
MB-11488 KV+XDCR System test: mem_used on each node reached to the bucket quota and rebalance hangs
- Closed