Excessive logging of canal daemonset


Whilst analysing cluster logs in Elasticsearch, I discovered that there are approximately 10 000 log entries per minute for my 3 node cluster. That’s a lot! 99% of these are from Rancher’s canal daemonset, virtually all of which are ‘INFO’ severity.

Looking at the environment variables for the canal daemonset, I see that this is by design…

  value: info

So my questions are:

  • Is it normal to see this many canal INFO messages for a 3 node cluster?
  • Can I simply change the value of FELIX_LOGSEVERITYSYS to ‘warn’?
  • Should ‘info’ really be the default?

Rancher: v2.0.5
Kubernetes: v10.1
Network Provider: Canal

This is fixed in v2.0.7

Good to know. Thanks for the reply.

I’ve rancher V2.3.2 and use calico-node image rancher/calico-node:v3.8.1, and I have still excessive logging on INFO level with lot of useless message. The verbosity has been fixed, or do we have to change log verbosity ?
2019-12-10 16:04:47.182 [INFO][48] int_dataplane.go 945: Applying dataplane updates
2019-12-10 16:04:42.377 [INFO][48] ipsets.go 306: Resyncing ipsets with dataplane. family=“inet”