Happy new year to you all xfs developers & users!
I'm evaluating xfs for use with the next incarnation of file server
(samba/linux) for my company, and while getting used to it (great
design, great implamentation - thanks SGI and xfs developers-) I came
across a few little problems:
-how does one recreate the log (external log, that is) after a device
(hdd) crash? the xfs partition resides on an raid5 device, but the log
itself resides on a simple disk; I browsed through man pages, but found
no clue of how to do it; does it work like that to just mount as usually
with "logdevice=/.." and automatically the log is recreated? sorry, I
haven't tried that, I'm two disks short now and a little away from the
server; fortunately, it's just a test setup (no data is lost).
-also, on an earlier test run, I created an xfs partition with an
internal log, (log) version 2; mkfs.xfs created it, but I could not
mount it, it complained something about "invalid sb" (not quite sure,
citing from memory); I'm on SuSE 8.1 with an updated kernel (still from
SuSE, ready compiled (from SLES8, used for oracle certification I
think), xfsutils and xfsdump from distribution). I know I should ask
them, but I have a feeling that I'm doing something wrong.
as a side note, when toying with xfsdump and xfsinvutil, I observed that
the UUID of the partition got set to 0000...; when doing a dump, xfsdump
gave me the message: xfsdump: getting UUID of the device: operation
(ioctl?) not supported (again from memory); yes, I shoud ask SuSE about
that, but does that impact restoring from back-up in any way?
thanks for your time,
dragos delcea
|