[DRBD-user] DRBD stuck in StandAlone

[Philipp Reisner]

Am Mittwoch, 17. März 2004 17:13 schrieb dhz_bf at walla.co.il:
> I have DRBD 0.6.11 that manages two partitions between two machines running
> kernel 2.4.18.  I am using heartbeat as a cluster manager, and it is
> configured to use the "datadisk" script to turn nodes to primary or
> secondary state as needed by the two nodes states.  I am using bonding to
> do HA on the NIC between eth0 and eth1. I have performed the following
> experimet.  I went to the switch to which both the primary and secondary
> nodes are connected and disabled the two links (eth0 and eth1) of the
> primary node.    Consequently, the secondary node moved to be Primary (and
> the other node Unknown).  Also, the cs became StandAlone. Then, I turned
> the two switch links back on.  As a result, the heartbeat returned the
> primary node to be primary and the secondary node to be secondary. 
> However, the cs of the secondary node remained as StandAlone, and the
> primary node's cs was WFConnection. There are pings between the two nodes;
> they communicate.  Let's not forget that the heartbeat detected that the
> two nodes communicate and ran the datadisk script on both nodes as
> required. Any idea why drbd on the two nodes don't communicate?  How can I
> recover from such a state?  Anything I can do to prevent it? Thanks,
>
> Doron Zifrony
>

Sounds a lot like the two devices were in Primary/Primary or they
tried to connect while they were Primary/Primary.

Then one device drops its network config.

-Philipp
-- 
: Dipl-Ing Philipp Reisner                      Tel +43-1-8178292-50 :
: LINBIT Information Technologies GmbH          Fax +43-1-8178292-82 :
: Schönbrunnerstr 244, 1120 Vienna, Austria    http://www.linbit.com :