0

I have a cluster with 6 nodes. As the official documents suggested, we need to change the replication factor of the system_auth keyspace to be the same as the number of nodes.

Now the system_auth seems corrupt because there are many version of system_auth, as the following: Different version of system_auth data

Some of the users could not be altered since I got null pointer exception: Alter user got Null Pointer Exception

I tried to use nodetool repair to fix it, but didn't help.

Could anyone tell me wha's wrong with my cluster and suggest me how to resolve the problem?

Thanks!

Andy Ho
  • 45
  • 8

1 Answers1

1

Did you try running repair with -pr for system_auth across the cluster? Using -pr requires it to be run on every node.

nodetool repair -pr system_auth

juniormint88
  • 206
  • 1
  • 3
  • Thank you for your reply. Actually, I followed this post to force Cassandra to rebuild the system_auth Keyspace. If I encounter this error in the future (I hope not :-) ), I will try this solution first and come back here to reply whether it works or not. https://stackoverflow.com/questions/25306785/how-to-reset-cassandra-superuser-when-cassandra-does-not-know-cassandra-defau – Andy Ho Nov 05 '17 at 22:33