[DRBD-user] Resync failures after reboot

[Andreas Pflug]

When rebooting a machine in a 3-way redundant drbd9.0.20-1 on Linux
5.0.21 (Proxmox 6.0-11), some resources fail to resync, either stuck as
"standalone" or at 97.3% or so.
Resource generation is done by linstor 1.2.0

I have:
hostD1 - secondary
hostD2 - secondary
hostD3 - secondary, being rebooted
hostN1 - diskless, primary and mildly loaded.

After a reboot, all active machines show hostD3 as "standalone" with
drbdadm status, while drbdadm on hostD3 shows "inconsistent" and hostD2
as standalone.

At that moment, hostD2 has network connections to hostD1 and hostN1 (as
expected), but doesn't listen on the designated port for that resource.
After drbdadm adjust, it will listen again.

Then drbdadm disconnect and connect --discard-my-data on hostd3 may
result in a finally synced disk again, or may sync until some high
percentage and stall there, requiring another disconnect/adjust/connect
round.

Does that ring any bells somewhere?

Regards,
Andreas