Summary: diskSuite quorum issue

From: Zaigui Wang <>
Date: Wed Oct 30 2002 - 02:29:16 EST
Wow, the response is overwhelming! I've also got
request for a summary. Thanks to:
Thomas M. Payerle
Scott Howard
Kevin Buterbaugh
Dan Astoorian
David Newton
Matt Harris
Paul Robertson
Chris Barnard
Richard Mackerras
Julie Peers
Manish Oberai

The majority response confirmed that there is a way of
disabling/loosening the >50% quorum requirement, which
is put this line in the /etc/system and reboot (for
DiskSuite 4.2.1 and newer):

set md:mirrored_root_flag=1

However, there is a reason why the quorum needs to be
greater than 50%. SDS works using MCA - Majority
Concensus Algorithm. When metadb disagrees on who is
updated and who should be sync'ed up, it looks at what
the majority of the db replicas say and decide for the
majority. If only half of the db replicas remain,
there is no majority and as a result, disksuite won't
be able to make a decision as to what is in sync and
what is out. Scott mentioned that "I've just checked
the source code, and it looks like it's pretty much a
first-in-best-dressed case if the metadb's are out of

IN a root-mirroring scenario, it won't be much of an
issue provided that one disk failed and not going to
be re-used with the metadb intact. But if you are
going to plug that disk back in and reboot the machine
with disabled quorum, you may be in for some
unpleasant surprises. 

One concensus is that since the machine shouldn't
crash due to a disk failure, after the failure is
detected, one can easily remove the the metadevice
replicas from the bad disk.  Then a reboot doesn't
have any issues.

Dan graciously provided this link for reference:

Paul provided the following:

HotJobs - Search new jobs daily now
sunmanagers mailing list
Received on Wed Oct 30 02:34:03 2002

This archive was generated by hypermail 2.1.8 : Thu Mar 03 2016 - 06:42:56 EST