Steve Hipwell
Steve Hipwell
@repeatedly the changes in that PR around UID and GID (actually GID wasn't set) appear to have been undone. I assume because the change of UID in entrypoint would need...
@repeatedly @edsiper the current behaviour on Alpine doesn't look correct, as the user's group doesn't seem to be set to `fluent`. I suspect that `-g` might have been used where...
@kenhys are you in a position to comment on https://github.com/fluent/fluentd-docker-image/issues/245#issuecomment-958818730?
Thanks @kenhys.
@kenhys now that the GID is being set how about making the UID & GID values absolute even if they're set to the current implicit values?
This continues to be an issue is there any more information on this?
@chancez there are a couple of examples of re-building this image as multi-arch you can find. My stevehipwell/fluentd-aggregator repo does it as a stop gap until this issue is addressed.
I suspect this is a race condition somewhere, my current thoughts are that it's the scheduler but I haven't had a chance to look at it further.
@gnufied the CSI driver looks to be doing everything correctly, AFAIK the only thing it needs to do is report the max PV attachments it can make. As reported above...
@gnufied I agree that the CSI driver is reporting correctly, which combined with the 60s wait fixing the issue makes me believe that this issue is actually happening elsewhere as...