[DRBD-user] Bringing up a failed master

Andreas Huck drbd at huck.it
Mon Nov 29 13:13:11 CET 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,

On Monday 29 November 2004 12:24, Gareth Ansell wrote:
> Andreas Huck wrote:
> 
> > your main server (drbd master) should become slave before
> > shutting down. Your cluster manager (heartbeat?) could handle 
> > this. It will come up as secondary again and the primary should sync the
> > changed blocks. But beware of any (misconfigured) autofailback feature
> > of your cluster manager.  
> > 
> > Regards,
> > Andreas
> 
> Thanks for the reply, we are only running drbd, not heartbeat, at the 
> moment.  Should I set the main server to secondary before I power it 
> down manually?
Hm, you might check your startup scripts then. Some sort of
clustermanager - at least the administrator or a script - is necessary.
Who tells drbd which server is primary?
If you use protocol C (and are cool enough ;-), simply unplug the
AC power cable on your primary. DRBD can deal with this, it's
a RAID-1. The application using your DRBD device should be able
to deal with this too.

> 
> Does drbd restart in the mode it was in previously, or does it always 
> start up in a secondary state?
Your clustermanager desides. On startup (modprobe drbd) sync is always
from newer to older - if drbd hasn't all information to decide, the
definition in "startup" of drbd.conf is cucial.

Regards,
Andreas

> 
> thanks
> 
> Gareth
> 
> 



More information about the drbd-user mailing list