On Mon, Oct 26, 2015 at 9:12 AM, sai krishnam raju potturi < pskraj...@gmail.com> wrote:
> hi; > we are working on a data backup and restore procedure to a new cluster. > We are following the datastax documentation. It mentions a step > > "Restore the SSTable files snapshotted from the old cluster onto the new > cluster using the same directories" > > > http://docs.datastax.com/en/cassandra/2.0/cassandra/operations/ops_snapshot_restore_new_cluster.html > > Could not find a mention about "SCHEMA" creation. Could somebody shed > some light on this. At what point do we create the "SCHEMA", of required. > 5.1) start the new node (as in step 6, with the correct initial_token list) 5.2) create schema 5.3) either run nodetool refresh OR restart the node again. I personally would just restart the node. Step 5 is also unclear : "Restore the SSTable files snapshotted from the old cluster onto the new cluster using the same directories." If you restore the system keyspace alongside your data keyspaces, they will contain a Cluster Name constraint. This Cluster Name should be different in the target cluster and the source cluster, otherwise there is a nonzero chance that they might talk to each other and that would be Very Bad. There is no supported way to change Cluster Name. For reasons such as this one, in general, IMO, one should not restore the system keyspace, but "nodetool snapshot" with no keyspace argument does snapshot it... The documents should also mention setting the conf files in the target cluster to have target cluster seed nodes, not the seeds from the old cluster. I have cc:ed docs AT datastax for their information, and I personally thank you for pointing out this omission. =Rob
