Rebalance exited with reason bad_replicas


#1

I did node down test.

  • sudo /etc/init.d/couchbase-server stop
  • and then …
  • after 30 secs, the node is auto-failovered.
  • And I did rebalance.

After rebalance, I show the below messages from log tab.

Rebalance exited with reason bad_replicas
ns_orchestrator002
ns_1@10.41.117.139
19:37:14 - Mon Jan 19, 2015

What does it mean, “Rebalance exited with reason bad_replicas” ?

==============
Test Environments

  • 8 nodes (128GB, all data is on memory)
  • data : 1.27 billion
  • 1 bucket
  • 1 replica
  • no auto compaction
  • OS : RedHat 2.6
  • Couchbase Server 3.0
  • Couchbase Client V2.0.3

#2

I’m not certain. Generally speaking, rebalance may have notice level messages if it runs into unexpected situations, but that doesn’t mean that it’d fail. @alkondratenko or @Aliaksey_Artamonau do you know what this message might mean?


#3

It means some replications that we expect to be active at the end of rebalance are not active. That is sometimes harmless (i.e. server will at least try to restore them). But it is indication that something is unexpected and might be point to some bug (like certain replications are not working at all). I recall 3.0.x had some bugs fixed in this area like: https://issues.couchbase.com/browse/MB-13069

Whether this is what your’re seeing or not I cannot say without logs


#4

@alkondratenko

Ok, I’ll share cblogs with u on jira u mentioned above.