Is this really a problem in Cassandra Cluster "2 replica were required but only 1 acknowledged the write"

0
votes

I recently setup 6 node C* cluster (C* version:3.0.13) and consistency-level QUORUM having RF=3. I keep seeing bunch of following exceptions.

Cassandra timeout during SIMPLE write query at consistency QUORUM (2 replica were required but only 1 acknowledged the write)

I understand i.e they will be written to other nodes eventually but Im still trying to find the root cause for the same.

Please shed some light about the same if it is okay to proceed OR anything that I should look after to improve code?

Highly Appreciate your response!

2
cassandracassandra-3.0
It's likely that your 2 other nodes are not connected. - ernest_k

2 Answers

0
votes

It seems your cassandra nodes are not doing gossip internally or some of the nodes not properly connected internally.. Please enable tracing in cqlsh and try the same. Also, run nodetool describecluster before running your query.

0
votes

Yes, it's definitely a problem. The writes will not be successful unless the required consistency is achieved.

In your case where you have RF:3, it sounds like the app is writing with a consistency of LOCAL_QUORUM so a minimum of 2 replicas must acknowledge the write for the operation to be successful.

This isn't a problem with your code. You need to investigate why nodes are unresponsive or down so not enough replicas are responding to the write requests. Cheers!