On Mar 5, 2012, at 11:49 PM, Ranjan Bagchi wrote: > it didn't kick the second shard out of the cluster. > > Any way to do this?
If you unload a core rather than just shut down the instance, that core will remove it's info from zookeeper. Currently, that won't make it forget about a logical shard I think though (just the physical shard). I think you have to then manually edit the zookeeper layout and remove the logical shard from under the /collections node. Feel free to file a JIRA issue around improving this if your use case requires it. - Mark Miller lucidimagination.com