Revision history [back]

click to hide/show revision 1
initial version

I have seen this on a Pike installation, there are a few things to check.

NTP, or time being out of sync can cause the volume service to die.

If you're running Ceph (your config suggests that are are not), then the calculation of RBD sizes looping through all volumes can also cause the service to die. In this instance, a cronjob to stagger restarts of the Cinder volume service is a workaround.