0

I've a Cassandra cluster on 456.456.456.23, I named it XYZ Cluster in yml file. While seed server coming up I get warning that the cluster name mismatch with another cluster ABC on 123.123.123.12 which is on completely different VM and should not be linked. How does it even know that there's another cluster exist! How do I solve this issue? I get this issue on Cassandra 2.x as well as 3.7 Appreciate your help...

Sïd
  • 65
  • 2
  • 11
  • I resolved this issue by completely purging the data from both the clusters and restarting the clusters. I still would like to know how to resolve these kind of issues without purging the data. – Sïd Jun 30 '16 at 20:36
  • https://wiki.apache.org/cassandra/FAQ#clustername_mismatch http://stackoverflow.com/questions/17162999/error-rename-cassandra-cluster-name-1-2-6 – undefined_variable Jul 01 '16 at 09:46

2 Answers2

1

This is related to unsafe removal of the node from existing Cassandra cluster. You haven't safely decommissioned and removed the 456.456.456.23 from your ABC cluster. Actually you are investigating it into wrong direction, its not 456.456.456.23 which is approaching or aware of ABC cluster. Its 123.123.123.12 which is aware and have information of node 456.456.456.23. So simply you have to check node details in your ABC cluster and remove its entry as node from ABC cluster. How?- Using below commands.

Check node tool details using below command on ABC cluster.

bin/nodetool status

It will give you output as below 

Datacenter: DC1
===============
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address      Load       Tokens       Owns    Host ID                               Rack
DN  456.456.456.23  ?          256          ?       08feba11-f04a-4226-8d08-c72c9e49ae25  r1

Datacenter: dc1
===============
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address      Load       Tokens       Owns    Host ID                               Rack
UN  123.123.123.12  261.35 MB  256          ?       3b95008a-0a67-4c3e-87f1-394fabfc0230  rack1

You can simply remove node by using below command use UUID to specify node. URL for help.

bin/nodetool removenode 08feba11-f04a-4226-8d08-c72c9e49ae25

Note-In production it was supposed to be decommissioned properly which should have avoided this issue to happen. you can use this URL to decommission node.

0

This can be related to what your seed is set to. If the seed IP is not set correctly in your yaml, everytime you start cassandra it will try to assign the wrong seed node. I would check both the yamls on the two clusters and ensure the IPs listed there are correct.

Ragit
  • 125
  • 1
  • 7
  • Seed was all set correctly. Don't know if at any time in the past seed was set incorrectly. Looked like it's some how picking up the wrong IP from somewhere. After data purge it stopped. – Sïd Jul 01 '16 at 17:24