On Mon, 29 Jan 1996, Rainer Toebbicke wrote:
>
> I guess what Steve meant was Solaris 'ODS' software RAID. Nothing to do with
> an external RAID controller where the system simply sees as a big disk.
>
> We use IBM RAIDs here, and of course we use Transarc's fsck (well,
> v3fshelper).
>
> What's interesting under Solaris is the possibility to just take a few disks
> and let the system define a RAID set on them. Since ODS has its own fsck, I
> wonder what happens if that one's replaced in turn by Transarc's fsck. A lot
> of smoke if you're sceptical...
Does ODS STILL have it's own fsck? I thought the fsck in 2.4 and later
actually understood journalling on its own. Could be wrong about that, I
haven't looked. In any case, though, the Transarc one doesnt. We've
been running some production servers with striped/concatenated/mirrored
disks with the Transarc vfsck for quite a while and it all seems okay.
Though Transarc has specifically told us they won't support that. But we
have so many big AFS volumes that having the bigger partitions is a big
win for us.
>
> BTW, Steve: there is a way to force Transarc's fsck *not* to fsck the logged
> file systems: I simply replaced the /dev/md/rdsk/dXX entries in /etc/vfstab
> by '-'. That way the meta-devices are not fsck'ed, the 'sync' done somewhere
> in /etc/init.d/SUNWmd.sync upon startup replays the log if all goes well and
> the file systems are 'clean' anyway. My worries are more whether this sort of
> magic is valid in all circumstances...
Yeah, that did occur to me - I think I'll try to take a look at the
transarc vfsck code and see if it could be easily modified to do the
"right" thing.
> > Cheers, Rainer >
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
> Rainer Toebbicke http://wwwcn1.cern.ch/~rtb -or- [EMAIL PROTECTED] O__
> European Laboratory for Particle Physics(CERN) - Geneva, Switzerland > |
> Phone: +41 22 767 8985 Fax: +41 22 767 7155 ( )\( )
>
>
>
Steve Hanson - FERMILAB, Batavia, Il.
[EMAIL PROTECTED]
