Details
-
Bug
-
Resolution: Fixed
-
Test Blocker
-
3.0
-
Security Level: Public
-
None
-
Centos 5.8 64-bit
Ubuntu 12.04 64-bit
-
Untriaged
-
-
Unknown
-
June 30 - July 18
Description
Tested on both OS, Centos 5.8 64-bit and Ubuntu 12.04 64-bit. Both test got the same issue as showing in steps below
Here is on Centos 5.8 64-bit
Install couchbase server 2.0.0 on 2 nodes
1:10.6.2.112
2:10.6.2.113
Create cluster 2.0.0 on those 2 nodes
Create a default bucket
Create 3 docs with 2 views per doc
Install couchbase server 3.0.0-890 on 2 nodes
3:10.3.3.224
4:10.3.121.241
Add node 224 and 241 into cluster 2.0.0
3.0.0 takes over master. Passed
Rebalance in 2 nodes 3.0.0-890 into cluster 2.0.0 passed.
Remove 2 old version nodes 2.0.0 nodes out of cluster.
Rebalance hang at around 50%
I don't see error like " Rescheduling backfill for vbucket 188 because backfill up to seqno 280 is needed but only up to 2 is persisted" in memcached log as in bug MB-11459
This bug may relate to bug MB-11176 that fixed view engine crashed.
Live cluster available