[mythtv-users] OT: A different LVM question

Jay R. Ashworth jra at baylink.com
Tue Jun 5 19:51:00 UTC 2007


I've been all around the horn on this for over a month, and my sister
is about to murder me.  Any help from experienced LVM hackers greatly
appreciated.

My sis's box has one big LVM group that contains 2 LV's spread across a
bunch of PV's.

One of the PVs lives on the boot device, a 200GB Seagate; it's 160GB.
Somewhere in the middle of that PV, there's a hole, which locks the
entire PATA controller up.  If you try to read those sectors, much less
write them, the OS can no longer even *see* the controller, much less
the drives.  Any of the drives.

That behaviour is repeatable one two different motherboards with two
different PATA controller chips.  It also happens both under SuSE with
a 2.6 kernel and with the built in disk drivers in Spinrite 6.  All 4
combinations.  I *assume* it's the drive adapter that's taking the host
adapters off line, because of that.

If I don't tickle the one bad area, everything's ok.

My plan was to use DD to strip off the head and tail of the volume,
onto a new 400GB we bought, and then do some juggling to clean up the
PV collection using pvmove -- pvmove was the application that initially
demonstrated the locking up, when it hit the bad spot.

But, I can't figure out how to get around the computer getting confused
because I will have two PVs with the same UUID.  Or because a desired
UUID PV is missing.

Anyone who *really* understands how LVM works who can give me any
insights into what's going (to go) on here?

Cheers,
-- jra
-- 
Jay R. Ashworth                                                jra at baylink.com
Designer                          Baylink                             RFC 2100
Ashworth & Associates        The Things I Think                        '87 e24
St Petersburg FL USA      http://baylink.pitas.com             +1 727 647 1274


More information about the mythtv-users mailing list