I am using Convoy-NFS with a NFS4 installed on an Ubuntu 14 instance. This is just for testing so I did not want to set up a gluster. Volume driver is convoy-nfs. Using the Rancher UI v 1.01.
Here is Scenario 1:
I create an Ubuntu container named Ubuntu1
- Volume mount = ubuntu1:/data1
Results: loads fine, volume mounts
I clone that container and call it Ubuntu2
- delete the volume mount and replace with ubuntu2:/data2
Results: Loads fine, but has two volumes mounted. ubuntu1:/data1 and ubuntu2:/data2
Here is Scenario 2:
Same as before for the first container
BUT when I clone it I set the Volume as ubuntu2:/data1.
Result: Container stuck at starting or In progress.
Why? because it is trying to mount 2 volumes to the same mount point. It has retained ubuntu1:/data1 from original and ubuntu2:/data1
It will never error out, The container can’t be deleted from the UI.
Docker stops responding. If I ssh to the instance and do a “docker -ps” it hangs.
In order to remove the container, I have to restart docker service, quickly do “docker ps -a” and docker rm
Furthermore any other instance that is trying to mount the same volume will hang as well.
I discovered all of this when I was trying to troubleshoot why my GitLab container (not the one from the catalog) will not mount its volumes. It was working fine. Then today I upgraded my underlying host and it moved to another host. Host is identical. Volumes will not mount.