Not sure if this is the right category for this. I’ve emailed support a couple of times, but haven’t gotten a response.
I’d like to change the way we do logging on a specific cluster/stack. I see that there is a Tutum-provided logrotate container that just nukes everything every 10 minutes.
I’d actually like to override that behavior.
How is the best way to have the default-logrotate container not run.
What I want to accomplish is have the logs hit the HD and then have Fluentd grab the logs from the log files (not stdin), filter and archive certain parts to S3 and then rotate the logs out. We need strong guarantees of log durability for some types of events.
It seemed the easiest way to do that was to create our own log-management container.
Some of our apps/stacks the standard logging setup is perfect. In some cases, we need stronger (and auditable) guarantees of things hitting the logs and getting archived.