Closed brndnmtthws closed 7 years ago
@brndnmtthws Were you ever able to make progress with this? We are seeing same error message.
I ended up just recreating the cluster. So no, I didn't make progress.
@brndnmtthws @erikriverson The fix we found is to do nodetool removenode <host_uuid>
. You can get the <host_uuid>
from nodetool status
.
@verma7 how do you use nodetool? I can't find anything in the DCOS documentation on this.
The code for this service has moved to a different repo.
Here's an example of invoking nodetool
from within the nodes themselves using dcos task exec
. In practice you'd probably want to do this from a separate client machine but this works in a pinch: https://github.com/mesosphere/dcos-commons/blob/master/frameworks/cassandra/tests/nodetool.py#L4
TLDR: You'd want to run something like dcos task exec TASK_ID bash -c 'JAVA_HOME=$(ls -d jdk*/jre/) apache-cassandra-*/bin/nodetool COMMAND'
I attempted to replace a node and got the following:
and: