[Bugs] [Bug 1462447] New: brick maintenance - no client reconnect

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1462447

            Bug ID: 1462447
           Summary: brick maintenance - no client reconnect
           Product: GlusterFS
           Version: 3.8
         Component: unclassified
          Assignee: bugs at gluster.org
          Reporter: mst at collogia.de
                CC: bugs at gluster.org



Description of problem:

We are using glusterfs in an OVirt environment. The setup might be a little bit
speical but here we go:

- 3 cluster storage nodes that have 2 bricks each to provide a distributed
replicated gluster filesystem

- a virtual IP that is managed by keepalived between the storage nodes

- several OVirt gluster clients that connect to the storage via native gluster
fuse mounts (initiallay over the virtual IP)

- Network infrastucture between all the nodes is IPoIB 

Version-Release number of selected component (if applicable):

Centos 7
glusterfs 3.8.12

How reproducible:

40-60%

Steps to Reproduce:
1. restart storage node for maintenance
2. wait for resync to complete

Actual results:
resync does not finish. New files are permanently added to the list of unsynced
filed. Looking at the client connections we see that the clients have not
reconnected to the node in maintenance.

Expected results:
client nodes should reconnect.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
You are the assignee for the bug.