-1

I deployed k8ssandra followed officail instruction on site https://docs-v2.k8ssandra.io/install/local/single-cluster-helm/.

After the k8ssandra deployment, I create a big table on cassandra DB. While I'm query the row nubmer by "select count(*) from TABLENAME", cqlsh always got ERROR like "Coordinator node timed out waiting for replica nodes' responses". So I wanna change some cassandra configurations. Could you tell me how could I do it or where the cassandra yaml file is for k8ssandra cluster?

Erick Ramirez
  • 13,964
  • 1
  • 18
  • 23
7072129
  • 29
  • 8

1 Answers1

1

This isn't a K8ssandra issue. The problem with your query is that an unbounded COUNT() does a full table scan.

This means that Cassandra has to read ALL the partitions from ALL the nodes in order to count them. There is no amount of tuning that will get around that. I've explained why in this post -- Why is COUNT() bad in Cassandra?.

As a friendly note, a reminder that Stack Overflow is for getting help with coding, algorithm, or programming language problems. For future reference, you should post DB admin/ops questions on DBA Stack Exchange. Cheers!

Erick Ramirez
  • 13,964
  • 1
  • 18
  • 23