SUMMARY Re: ufsdump of Solstice metatrans device

David Thorburn-Gundlach (dtg@cae091.ed.ray.com)
Mon, 22 Sep 1997 11:46:22 -0400

Hi, folks --

David Thorburn-Gundlach wrote:
...
> Here's the rub: when we back up the mirrored, striped partition (a la
> "ufsdump 0f srv:/tape /dev/md/dsk/d60"), we don't get all of the data.
> Strangely enough, there are no errors, and what does get backed up does
> not seem to be in any particular order (ie, oldest or newest or
> alphabetical).
...

Unfortunately, I didn't get any concrete answers to this one; I guess I
do a good job of playing "stump the experts" :-)

Many thanks to
- Rich Kulawiec
- David Wolfskill
- Arthur Darren Dunham
- Glenn Satchell
for their suggestions. I haven't yet been able to look at a listing on
tape compared to a listing of the disk itself to compare inode numbers,
but most folks suggested that 2G is the limit and that ufsduump will
pick up inodes pretty much in order. I have, in the past, personally
dumped over 20 (twenty) G with ufsdump with no problem, though...

Because we're already using Legato here for another set of machines and
they have no problem dumping large metadevices, we're going to set up
these servers as Legato clients as well (sorry, Rich ;-) and send them
to our 1T DLT stacker (a nice added bonus!). Nonetheless, we sure would
like to know why the problem exists, because we will almost certainly
want to ufsdump the Legato server and its database disk (which will
probably be a large metadevice)...

:-D

-- 
David Thorburn-Gundlach
dtg@cae091.ed.ray.com,david@bae.uga.edu
Raytheon  508/440-2016 or 508/440-2317