On Tue, Sep 10, 2013 at 11:03 AM, Langston, Jim <jim.langs...@compuware.com>wrote:
> http://comments.gmane.org/gmane.comp.db.cassandra.user/29753 > For step 3 in the instructions, I moved LocationInfo located in the system > keyspace to another > directory and when I try to restart the node, the directory is re-created, > but still get the error. > As of 1.2.x, this information is now kept in a different place. It is now kept in a column called "cluster_name" in a row with key "local" in a CF called "local" in the keyspace "system". You have two options : 1) change this key on all nodes 2) do a rolling restart 3) realize you will have two partial clusters until you complete your rolling restart, with attendant consequences This should work because the cluster name check only occurs at node start time. ... or you can ... 1) stop all nodes 2) move the system keyspace aside entirely 3) start nodes in such a way that your cluster re-coalesces (start seeds first, etc.) 4) reload your schema The latter is probably what I'd do. If you do either, please respond on thread and let us know your results. =Rob PS - Prompted by this thread and my dissatisfaction with either of the above workarounds, I have created the following feature request : https://issues.apache.org/jira/browse/CASSANDRA-5997 : -D option to change cluster name
