[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ogfs-dev]Finding some of the trouble spots with OpenDLM

Hi all,

Between yesterday and today, I've found some of the trouble spots that
have kept us from mounting successfully, and some that would cause us to
lose lock records.

Mount problems -- deadman algo is not working right.  While documenting,
we learned some new info about how OpenDLM works.  Stan's rethinking the
deadman algo.

Lock loss -- dlm_purge was being called inappropriately in 2 spots
(locking with NOEXP, and when journal recovery was done).  We need to
figure out how to handle orphan locks, after a node dies.

I just checked some files into the src/locking/modules/opendlm directory

-- add/edit some comments
-- add/edit some debug output
-- hack out some code that was causing troubles

THE HACK IS A HACK.  It is bypassing some of the recovery features while
we figure out how to do this properly.

I'm now consistently *successful* in mounting the first (1st in time,
that is, not cluster configuration) node, and am making progress on
mounting the second node.  Stan's also working on this.

-- Ben --

SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
Opengfs-devel mailing list

[Index of Archives]     [Kernel]     [Security]     [Bugtraq]     [Photo]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Clusters]     [Linux RAID]     [Yosemite Hiking]     [Linux Resources]

Powered by Linux