external-dns: Log messages showing as errors in Google Stackdriver

I would like to reopen issue #772 because our Stackdriver logs are spammed with errors that are actually external-dns logs with level=info. There are a couple ways we can fix that:

  1. Provide a configuration option to write log severity in severity JSON attributes instead of level. Eg. by setting --log-format=stackdriver_json.

  2. Provide a configuration option to write all debug and info logs to stdout, and only rest to stderr. Or list of severities that would be sent to stdout.

I think the first option is preferred because then we can use severity filters in Stackdriver. Please note that there is no way how we can fix Stackdriver configuration to recognize info level because it’s Fluentd agent configuration is managed by GKE team 😕.

About this issue

  • Original URL
  • State: closed
  • Created 4 years ago
  • Reactions: 8
  • Comments: 24

Most upvoted comments

@AndrewDryga Ran across this myself in GCP running on GKE/GCP Monitoring (Stackdriver). I decided to try changing the log format to json and stackdriver appears to be categorizing the messages appropriately now, see attached screenshot.

I’m using the bitnami external DNS chart CHART NAME: external-dns CHART VERSION: 6.11.3 APP VERSION: 0.13.1

with the following in my values.yaml file logFormat: json

You can see the errors switch to info’s on redeploy of the chart.

external-dns-stackdriver

@Saeger it doesn’t solve the issue for stackdriver, our format was set to JSON for a while already.