Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Wed, Jan 19, 2005 at 06:10:54PM -0800, Jeff Buck wrote:
> > I had this set up, but could not get it to fail over. One physical
> > volume just couldn't be recognized by the secondary. LVM could find
> > it on the underlying SCSI devices, though. I'm not sure what the
> > problem was.
>
> You really can't have lvm2 finding the the lvm volume on the underlying
> scsi device. LVM2 must be configured on both boxes to prevent it from
> finding anything on the scsi device. The scsi device and the drbd device
> are not identical, but lvm2 will be perfectly inclined to believe that
> they are. We don't use any automated failover systems, but we still had
> tons of problems when the boxes could find lvm on the scsi devices. You
> need to change lvm.conf to filter out the scsi devices, but beware, the
> filter examples that came with our fedora core 3 box, were very bad, and
> mostly incorrect. Test your filters well to make sure they're working
> properly.
Jeff,
Thanks for the message. I agree it's basically impossible to
configure lvm on top of drbd with an improper filter. Like you,
I learned about this the hard way.
When running on drbd, I had it filtered to only use drbd devices --
though I see that this looks like the problem from my original
question! My apologies for the confusion.
Also, while I'll eventually setup heartbeat, right now I'm just
trying to establish the ability to fail over manually -- probably
similar to the way you're doing it.
The original devices section in lvm.conf was:
devices {
types = [ "drbd", 32 ]
filter = [ "a/drbd.*/", "r/.*/" ]
}
Using this setup, after drbd failover, one physical volume was not
found when running vgscan. The error was:
mason4 resource.d # vgscan
Reading all physical volumes. This may take a while...
Couldn't find device with uuid
'PU25qj-ayHw-vneG-zQmr-WWRV-tuRW-HF51b1'.
Couldn't find all physical volumes for volume group vg.
Volume group "vg" not found
The original primary can find it just fine when I fail drbd
back over.
When I do that, and remove the one offending physical volume,
the other two are found just fine by both machines.
The two physical volumes that work are 1.0 TB each. The one that
didn't work was 1.3 TB. However the latter had been my original
test device, and it worked before I added the other two devices.
The really weird thing was that when I shut down drbd and removed
the filter from lvm.conf, the secondary machine could find all three
physical volumes and lvm worked just fine. This despite the fact
that everything had been configured on the primary machine via drbd on
the raw devices.
Thanks again for the response! Any further help would be greatly
appreciated.
-- Trey
--
Trey Palmer Systems Development Engineer trey at isye.gatech.edu
Georgia Tech Industrial and Systems Engineering 404-385-3080