[3160] in Release_7.7_team
Linux TSM backup
daemon@ATHENA.MIT.EDU (Alex T Prengel)
Mon Feb 25 14:15:57 2002
Message-Id: <200202251915.OAA17225@dit.mit.edu>
To: release-team@MIT.EDU
cc: alexp@MIT.EDU, ops@MIT.EDU
Date: Mon, 25 Feb 2002 14:15:53 -0500
From: Alex T Prengel <alexp@MIT.EDU>
It's looking like using the TSM "Virtualmountpoint" directive will work
around the "TSM won't back up ext3 filesystems" problem, at least when
using the dsmc command-line client and the scheduled backup daemon.
Unless amb has a better suggestion, we'll most likely recommend this to
users tomorrow or the next day, after a little more testing (we need to
get back to users quickly on this one).
Also- as a by-product of all this, I learned that tytso is now working
in IBM's Linux Technology Center. This could come in handy... he also
tells us, about the problem:
>I assume that TSM/ADSM is actually opening the raw device to do its
>backups, and it's getting tripped up by the the has_journal capability
>flag? Or is it OK for the filesystem to have has_journal capability
>flag, as long as it is *mounted* as an ext2 filesystem? I strongly
>suspect the former, but I'd like to make sure it's confirmed.
>
>You could test this by taking a filesystem with a journal --- i.e.,
>which has had "tune2fs -j /dev/hdXXX" run on it, mounting it as an
>ext2 filesystem (mount -t ext2 /dev/hdXXX /mnt), and then checking to
>see if TSM has heartburn over it. Then after you unmount the
>filesystem, run "tune2fs -O ^has_journal /dev/hdXXX", and then
>remounting the filesystem as ext2, and seeing that TSM is fine with
>it.
If someone has a Linux machine with a TSM account, can you check this
out? Thnx-
Alex
