[DRBD-user] Using discard-zero-changes and consensus with fencing

Lars Ellenberg lars.ellenberg at linbit.com
Thu Feb 26 16:39:30 CET 2009

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


On Thu, Feb 26, 2009 at 03:59:44PM +0100, Federico Simoncelli wrote:
> On Thu, Feb 26, 2009 at 2:33 PM, Lars Ellenberg
> <lars.ellenberg at linbit.com> wrote:
> >> Is my analysis correct? Is there
> >> something wrong in my configuration?
> >
> > you have _two primaries_
> > the new UUID is created when connection is lost.
> >
> > when the connection is first re-established,
> > they are both still Primary, they have diverging data sets:
> > => split brain is detected.
> >
> > after-sb-2pri would be the setting that is used here
> > (_2_ primaries, when split brain is detected)
> 
> I don't think the after-sb-2pri would be the setting used in this
> situation. After the fencing, node2 boots again and the "drbdadm
> adjust" and "wait-con-int" (in the init script) doesn't make it
> primary until the "sh-b-pri" command which is issued later on. Am I
> missing something?

maybe I am missing something.
you did not include full logs,
especially during the handshake after the reboot,
where the split brain is detected.

apparently they are not able to find "zero changes" or "consensus",
and thus decide to "disconnect".

at least that is what the provided log says.

-- 
: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com

DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
__
please don't Cc me, but send to list   --   I'm subscribed



More information about the drbd-user mailing list