Cassandra cluster restart

0
votes

I'm building two node Cassandra 2.1.2 cluster (GossipingPropertyFileSnitch) with following sequence:

  • start seed
  • start node

Now "nodetool status" shows both nodes UN on both machines.

Next: - stop node - stop seed

  • start seed "nodetool status" on seed shows seed UN, but no second node at all (I expected node shown as DN).

On seed create keyspace with replication factor 2 (strategy doesn't matter), create table, insert something into table.

Run node, now "nodetool status" shows both nodes UN on both machines again.

Now created keyspace and table are seen on node (create was propagated from seed), but is empty from node POV.

node rebuild or repair fix this, but actually I expected synchronization will be done automatically after node start.

Thanks.

P.S. It seems that 2.0.11 works fine.

1
cassandra

1 Answers

0
votes

This doesn't seem correct. Cassandra will not automatically remove a node from a cluster just because it is down. You would need to run nodetool removenode or nodetool decommission.

So after the sequence: start seed, start node (both shown as UN in the same cluster), stop node, stop seed, start seed, you would see UN for seed and DN for node.