0

We have Cassandra set up with 3 datacenter and 17 nodes in each data center. We are seeing strange issue as on one node in DC1 is acting always as coordinator node and if for some reason this node goes down whole cluster is not reachable. We are seeing this very frequently and we have not mentioned this IP in our code base.

Cassandra : 2.0
Astyananx : 3.8
cassandra.astyanax.connectionPoolType=TOKEN_AWARE
cassandra.astyanax.defaultReadConsistencyLevel=CL_LOCAL_QUORUM
cassandra.astyanax.defaultWriteConsistencyLevel=CL_LOCAL_QUORUM
cassandra.astyanax.retryCount=3
cassandra.astyanax.discoveryDelayInSeconds=60
cassandra.astyanax.discoveryType=RING_DESCRIBE
cassandra.astyanax.cqlVersion=3.0.0
cassandra.astyanax.targetCassandraVersion=2.0

Please help us here.

Thanks Virendra Agarwal

viren
  • 1,073
  • 1
  • 17
  • 41

1 Answers1

0

As your replication factor one and using connectionPoolType TOKEN_AWARE, it should attempt to contact only the nodes that actually store the data and in your case it is 1 node only in local dc. Check this for more details.

Laxmikant
  • 1,551
  • 1
  • 13
  • 30
  • Thanks i saw that but problem is when this one node is down application wont connect to any other node. Which is not a default behavior If node is down other node should be connected as coordinator and writes should work. I am assuming this should be behavior. – viren Oct 12 '18 at 07:37
  • have you tried ROUND ROBIN ..just to narrow down the issue ? – Laxmikant Oct 12 '18 at 11:02