[DRBD-user] DRBD 9.0.7-1 wierd issue

Jaz Khan jazmphone at gmail.com
Sat May 20 22:08:32 CEST 2017


Hello Linbit team,

I am running DRBD 9.0.7-1 on 3 physical machines running on CentOS 7.3
This is what happened with me two times in a month.

I have 3 machines: Node-A (primary), Node-B (secondary), Node-C (secondary)

Everything was setup using drbdmanage tool and all 3 nodes were perfectly
up-to-date until I powered Off & On my Node-B. Below is the result.

On Node-A (primary):
# drbdadm status
Node-B UpToDate
Node-C UptoDate

On Node-C (secondary):
# drbdadm status
Node-A UpToDate
Node-B UpToDate

On Node-B (secondary):
# drbdadm status
Node-A UpToDate
Node-C Outdated   <<<<<<< This is what happened with me twice.


Why Node-B is seeing Node-C "outdated" but Node-B is seeing it UpToDate? I
have recreated Node-B twice but whenever one of my secondary goes down this
problem appears and it is so much painful to recreate Node-B from scratch
because drbdmanage is still very immature and does not work well with "dbus"

When I issue this command "drbdmanage list-nodes" from Node-B it shows both
A & C status Offline. But when I issue the same command on A & C it shows
"Ok" status for everyone.

When I try to remove Node-B from Node-A which is primary, it deletes it but
the status from "drbdmanage list-nodes" shows *pending actions: remove" and
it is stuck there forever.

Unfortunately, these 3 servers are running in production.

*If any Linbit & DRBD maintenance or dev team could rescue me from this
issue, I would really grateful.*

Best regards,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20170521/528127f3/attachment.html>


More information about the drbd-user mailing list