Connection to cluster timeout after adding node.

[prathamesh saraf]

In my application i am running a single node cassandra instance on one machine and I insert data into cassandra from another machine.

While the inserts are being performed,I add another node to the cluster. The new nodes comes up and my is successfully added to the cluster. I receive this error message a few time and then once the table is created it is no longer received

UnknownColumnFamilyException reading from socket; closing
org.apache.cassandra.db.UnknownColumnFamilyException: Couldn't find table for cfId 7c999bd1-7ac0-11e6-bf17-39ce1d940986. If a table was just created, this is likely due to the schema not being fully propagated. Please wait for schema agreement on table creation.

After this my inserts no longer start working and i loose the connection to the cluster. The first node i was using to connect to the cluster no longer accepts connections even through the CQL client.

Do i need to reinitialize the session object for the driver or am I missing something. The same scenario works when there is no data being pumped during node joining.

Any help us highly appreciated.

[Michael Penick]

Does "CQL client" mean cqlsh? If cqlsh is unable to connect it's likely a Cassandra issue.

You don't need to reinitialize the session object. The session will automatically reconnect to nodes.

Mike

--
You received this message because you are subscribed to the Google Groups "DataStax C++ Driver for Apache Cassandra User Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cpp-driver-user+unsubscribe@lists.datastax.com.

[prathamesh saraf]

> To unsubscribe from this group and stop receiving emails from it, send an email to cpp-driver-us...@lists.datastax.com.

Yes the CQL client is the cqlsh.. The connection is restored after a while, but i loose my inserts during that time frame. I believe that cassandra nodes could be added during data operations (Read/write) being performed.

Also in the driver, can I set the processor affinity for the threads? Running cassandra-stress tool gets me good insert speeds and it runs at 200 threads "Running write operation with 200 threads". I tried setting the cass_cluster_set_num_threads_io o a large number like 128, but i din receive the required performance. My CPU utilization with cassandra-stress spikes at 90% per core. This same is not happening with c++ driver.