/var/log/ on nodes is being polluted over time, and it’s root partition

[Mike Scherbakov]

Not all logs are being rotated to master node, and we still have increasing of /var/ in size over time. It means that without periodical clean up, we can face a situation when node gets stuck. In case with controller it would mean stopping of operability of the whole OpenStack environment, as our HA architecture doesn’t yet allow to handle such kind of situation, and failover won’t happen. In case with compute node, it may lead to deniel in hosting of new instances. I’m not sure if it can influence already running kvm processes.

Let’s think what would the best approach here, should we have only /var/log/ LVM? Should we enable periodical clean up by cron?

[Bogdan Dobrelya (bogdando)]

I believe we should address all Logs Storage considerations including LVM ones. See https://blueprints.launchpad.net/fuel/+spec/manage-logs-with-free-space-consideration

[Bogdan Dobrelya (bogdando)]

(https://etherpad.openstack.org/p/manage-logs-with-free-space-consideration)

[Vladimir Kuklin (vkuklin)]

this bug requires logs managing blueprint

[Bogdan Dobrelya (bogdando)]

As far as related BP targeted to 6.0 release we should retarget this issue as well

[Bogdan Dobrelya (bogdando)]

superceeded by https://blueprints.launchpad.net/fuel/+spec/manage-logs-with-free-space-consideration