[DRBD-user] Current Primary shall become sync TARGET

Steve Purkis steve.purkis at multimap.com
Thu Sep 30 16:07:36 CEST 2004

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.

Hi Luca,

I've experienced the same 'problem' -- see 

In summary - it's not a problem, the problem is that you do not 'fence 
off' your IO requests, so even if heartbeat shuts down drbd on your 
original primary, it still writes to disk as it shuts down -- this may 
be in the form of a timestamp or similar when you umount the disk, or 
something else.

The solution is to reboot after a network failure.  STONITH may help 
Otherwise, do a manual reboot.
Alternatively, you might find Mon useful, I haven't yet tried setting 
it up to kill DRBD (the idea that this would be the same as IO 


On Sep 30, 2004, at 14:46, Luca Uburti wrote:

> This is my first post to this ML, I hope this problem
> has not been already discussed here, but I couldn't
> find a solution elsewhere.
> I installed drbd 0.7.4 + heartbeat rpm's for RH8 from
> UltraMonkey + ipfail.
> I set up a cluster of two computers: node1 (primary)
> and node2 (secondary), sharing the drbd device on the
> net with Samba. Each node has two Nic's, one for the
> data and one for the heartbeats on a crossover cable.
> Everything seems to work fine except for one thing.
> Whenever I disconnect the eth cable from node1 (the
> "data" cable, not the heartbeat cable), node1 becomes
> secondary and node2 becomes primary, BUT if I later
> reconnect this cable, the node2 prints this message in
> the logs:
> "Current Primary shall become sync TARGET! Aborting to
> prevent data corruption."
> and then goes to StandAlone mode, while node2 stays
> secondary/WFConnection forever.
> If I reboot or I just pull the power cord from a
> computer (primary or secondary), the cluster seems to
> react normally. It fails over / fails back without a
> glitch.
> The problem is ONLY when I disconnect the primary's
> ethernet cable and then connect it back.
> Changing the auto_failback option doesn't solve the
> problem.
> I wonder if this behaviour is by design or if there's
> something wrong with my cluster configuration.
> ___________________________________
> Scopri Mister Yahoo! - il fantatorneo sul calcio di Yahoo! Sport
> http://it.seriea.fantasysports.yahoo.com/
> _______________________________________________
> drbd-user mailing list
> drbd-user at lists.linbit.com
> http://lists.linbit.com/mailman/listinfo/drbd-user

More information about the drbd-user mailing list