Created attachment 703346 [details] Adding vdsm,glusterfs logs of both the nodes and engine.log Description of problem: When doing a force removal of the server it is not detaching the server. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. Create a new cluster. 2. Add two hosts into the cluster. 3. Create a volume on one of the hosts in the cluster and start it. 4. Now put the host in maintenance mode. 5. Do a force removal of the host. Actual results: The hosts gets removed from GUI but the volume is still present and host is not getting the detached from the CLI, i.e when you do a peer status it still shows the other peer and the volume info as well. Expected results: The hosts and the volumes created should be removed and the server should get detached from both CLI and GUI. i.e. both the UI and CLI should be sycned. Additional info:
This should happen only if there was no other UP server in the cluster when you tried to force remove this one. Can you please confirm that? The logic is like this: - When "force" checkbox is not selected Will perform "gluster peer detach" from one of the UP servers Will fail if there is no UP server in the cluster - When "Force" checkbox is selected Will perform "gluster peer detach force" from one of the UP servers Will simply remove the server from DB if there is no UP server
Created attachment 762622 [details] Build bb3
This bug continues to be an issue in build bb3 - see attached log files.
If there's an UP server in cluster, force remove of server will detach server from cluster. However, if server has associated bricks, force remove is not permitted on server. (unless all servers in cluster are in Maintenance mode)
As per discussion, the Console behaves as expected. Documentation to be updated to detail the workflow to be performed when removing server from CLI for it to be in sync with the Console Adding to tracker doc bug 973888