I'm running a project on GKE and fluentd pod is stuck in a bad state. I accidentally flooded it with log entries and it cannot recover.
The pod is up to its memory limit and fluentd is being restarted every 12 seconds. the logs look like this repeatedly:
2017-03-16 03:55:26 +0000 [error]: fluentd main process died unexpectedly. restarting.
2017-03-16 03:55:27 +0000 [warn]: 'block' action stops input process until the buffer full is resolved. Check your pipeline this action is fit or not
2017-03-16 03:55:27 +0000 [warn]: 'block' action stops input process until the buffer full is resolved. Check your pipeline this action is fit or not
2017-03-16 03:55:38 +0000 [error]: fluentd main process died unexpectedly. restarting.
I have tried:
- Deleting the pod.
- Downgrading and upgrading the k8s version on the node.
- Looking in host /var/log/fluentd-buffers for massive files
None of which helped
Is there a way to force the fluentd in the container to flush work again?
It seems like a bug that it can even get into this state.
Thanks
p