plugin volume not found on inspect or rm

[RemyCH]

Hello,

We use your plugin on different dockerEE clusters with different configurations : UCP 2.1.5 Docker engine 17.03 Plugin : 17.07

UCP 2.2.4 Docker engine 17.06 Plugin : 17.07

UCP 2.2.4 Docker engine 17.06 Plugin : 17.10.0

UCP 2.2.5 (release candidate) Docker engine 17.06 Plugin : 17.10.1

It seems like we have systematically the below issue with your plugin (any version 17.07, 17.10.0, 17.10.1) under 17.06/UCP 2.2.X.

Here is the issue : Using admin user, or lambda users, we can create a volume. Then we see the volume via "docker volume ls" But using bundle, or UI, we can't inspect or remove this volume. It returns a "volume not found".

• manually via bundle :

  bash-4.2$ docker volume ls |grep remytest
  netapp:latest       remytest
  netapp:latest       remytest
  netapp:latest       remytest
  netapp:latest       remytest
  netapp:latest       remytest
  netapp:latest       remytest
  netapp:latest       remytest
  bash-4.2$ docker volume inspect remytest
  []
  Error: No such volume: remytest
  bash-4.2$

• via UI :

  We can inspect the volume without using the ucp bundle. Informations about my volume in /var/lib/docker/volumes/metadata.db are there :

strings metadata.db |grep remytest remytest{"Name":"remytest","Driver":"netapp:latest","Labels":{"com.docker.swarm.whitelists":"[\"node==dckdevdck181\"]","com.docker.ucp.access.label":"/Node/Workload/TEAM-DEV/mydev","com.docker.ucp.collection":"6ea02707-e2ff-4a3a-bfac-39464570cd73","com.docker.ucp.collection.6ea02707-e2ff-4a3a-bfac-39464570cd73":"true","com.docker.ucp.collection.95227568-0f22-42bd-a90d-b331de8097aa":"true","com.docker.ucp.collection.af70838e-f5d8-4a79-92e2-c2400d3e5b82":"true","com.docker.ucp.collection.b5a19ceb-6b67-485e-a250-2988585ba7e5":"true","com.docker.ucp.collection.root":"true","com.docker.ucp.collection.swarm":"true"},"Options":{}}

It turned the plugin in debug mode, but when we inspect the volume : it returns strictly nothing in docker engine, or netapp plugin logs.

Do you have any idea?

Thx Remy

[kevtainer]

we've observed this behavior as well. the only workarounds we've found is to open a shell session to a manager node to remove the volume, or running a container on the manager node with access to the host docker socket.

[RemyCH]

Docker should fix it soon. I just checked. So i close there.

[lordgnu]

@RemyCH Can you please leave a link to what you were looking at in the comments? This issues affects me several times a week...

[adkerr]

@RemyCH I too would like to know where you found this confirmation, thanks!

[RemyCH]

Internal case @docker.