CPU Consuming in particular node


Hi Team,

We have 3 nodes cluster with AZURE load balancer setup (v19.2) and we are getting CPU issue on particular node only. how to find the issue and how to set equal load to all the nodes .


Hi Raj, can you confirm that the load balancer is distributing traffic evenly across the nodes? Take a look at the SQL dashboard and confirm that each node is seeing approximately the same number of queries per second. Let me know what you find with that first!

HI Rafi,

Sorry for the delay responses.

I have checked cluster level around 60 to 70 connection per sec .
Each nodes getting 20 to 25 connection.


Hi @rafiss ,

We are getting “compaction_time_cpu_micros” alert in log files.

what’s compaction in CRDB !


Hi @Raj, compaction is a background job that happens within the underlying storage engine used by CockroachDB. There is a good overview of it here: Leveled Compaction · facebook/rocksdb Wiki · GitHub

Which version of CockroachDB are you using? Versions 19.2 and earlier only support RocksDB as the storage engine, and in versions 20.1 and later support for Pebble was added.

I believe older versions might have a few known issues with the CPU usage of compaction, so knowing the specific version you are on will help. Also, if you are able to try the latest stable release, I’m curious how the performance changes.

1 Like