Open msuringa opened 8 years ago
I'll have to look into this - I would think that docker daemon would send the proper command to the volume plugin when terminating any running containers. If daemon gets killed and can't cleanup, then I might expect to need to do some manual cleanups.
What version of docker ? What OS? Is it a clean shutdown or an unexpected termination?
OS is RedHat 7.2 Docker version: 1.9.1 I see this behaviour when I do a clean restart/stop of the docker daemon.
I have resolved this issue locally by adding some logic to check if the image that's being mounted has watchers anywhere, if it doesn't, it will cancel whatever lock is in place and lock it for the local server.
When I restart the docker daemon which has containers running that use this volume driver, the running containers get stopped/killed, but the all mounts remain locked. Is this something the volume plugin can resolve, or is this handled in the docker daemon process?