Unintentionally we have lost (terminated in AWS console for a manual error) 5/12 Cassandra nodes from a AWS datacenter. So we are trying to get new nodes to replace them after "nodetool removenode" on dead nodes.
Can I add new servers to existing datacenter with "auto_bootstrap: false" and run nodetool rebuild on new nodes?
And Can I run nodetool rebuild on multiple nodes at a time? if yes, what could be the downside of it?
Thanks in Advance!
I just got to know that, yes, we can do. Normally this is how to rebuild a whole DC but we can also use it to rebuild just a subset of nodes.
@vkayanala_42513 you can but when you disable auto bootstrap, the node will automatically join the cluster without any data on it and it will accept reads from the application (without data to return) so will cause you issues. It is a bad idea to do this unless you are adding nodes in a DC that is not serving traffic to the application. If you allow a node to bootstrap normally, it won't accept any read requests until the bootstrap has completed. Cheers!
@Erick Ramirez This is what I did.
- Add new nodes to cluster with auto_bootstrap false in cassandra.yaml. - Perform a nodetool rebuild pointing to the other DC in the cluster. - repairing old nodes in cluster.
Cheers!
4 People are following this question.
