0

We have a 3 node Cassandra cluster with a small amount of data on 'em (about 2.3GB total). The nodes are hosted on GCE and each have four cores and 16GB memory. We're running the Debian package build of Cassandra 2.1.1. The lone keyspace we have (besides system) is configured with SimpleStrategy, replication factor 3.

We have monitoring set up to report on schema disagreement, using nodetool describecluster. The monitoring checks each node for disagreement about once a minute, and we see a report once or twice a minute. The system seems to automatically deal with the problem, however. I have yet to see the problem when I manually run describecluster.

What should we do to avoid this constant flapping? Is it representative of a serious problem? I haven't tried much to solve the problem as I have no idea where to start. I've upgraded the servers a few times, adding cores and memory, but that does not seem to have had any effect on this particular issue.

Improve this question

1 Answer 1

Reset to default
0

If you are adding/removing keyspaces or altering the schema, then you are going to have disagreements for a brief period. You probably want to change your monitoring to only alert if the mismatch lasts for more then 5 minutes or something.

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .