How to start CDB without INFO or WARNING logs

I have 11GB of logs in one of my machines. The documentation says I can either have logs turned on, or off. How do I start CDB such that it only spits out ERROR or FATAL logs?

Hi kae, thanks for your question.
If you choose to enable automatic logging to files, as is the default, then all the log messages will be written to the log files; however the log files are split over time and you can remove older files.

If you redirect logging output to standard error with --logtostderr, in the last beta release you could not filter out messages based on logging level, but in the next beta you will be able to do so using the --alsologtostderr flag, e.g. --alsolotostderr=ERROR.

Does this address your concern?

I don’t think so.

“–alsologtostderr” implies that the standard log files will be created as before, but now there are even more logs, this time in STDERR.

what I’m looking for is a way to not have the INFO and WARNING logs write at all.

I don’t think it’s right to have to go in and periodically clean out files I never need/want in the first place, on every single server in the cluster.

Thanks for your feedback. I have filed your request here: https://github.com/cockroachdb/cockroach/issues/11654 - feel free to add a comment there with any suggestions you may have about how you would like a fix to look like from your perspective. If you have an example/specific command-line interface in mind, let us know.

As a stop-gap, until we fix the issue, I could invite you to consider the logrotate utility: https://linux.die.net/man/8/logrotate.

Best

1 Like