[DRBD-user] "drbdadm down" takes a long time to complete

[d tbsky]

Janusz Jaskiewicz <janusz.jaskiewicz at gmail.com>
>
> Hello.
>
> I have a 3 node drbd cluster, where 2 nodes (prod01, prod02) are managed by pacemaker and the third node (dr01) runs raw drbd.
>
> The replication works well, but sometimes I have an intermittent problem with bringing the drbd down.
> It happens regardless if the node is managed by a pacemaker or not.
> The problem is that the drbdadm down <resource> command takes a very long time to complete (a couple of minutes).

I've got the same problem.
I had used drbd since 8.4 for many years. I never get any problems
about "drbdadm down".
recently I upgraded the OS to rhel 8.5 and drbd 9.1.4. when I try to
shutdown the pacemaker cluster, I notice one drbd resource hang there
and  "drbdadm down" manually hang too.
I think maybe I can reproduce it. but I want to know what to check
when it happens...