Sun 280 - boot failure - mirrored drives

[GallopingGhost]

Have a Sun 280 running Solaris 8 with mirrored drive. During a routine reboot the /var fs is giving a SUPERBLOCK error. Ran the fsck command against /dev/md/rdsk/d5 and used the -o b=# switch with 32, 4096, and 8192 values and system is still down. Keeps telling me to use another SUPERBLOCK value.

I think I need to unlink the d5 from its mirrors which are d6 & d7 (came from the command metastat -i d5 ) and then run the fsck command. Am I heading in the right direction and what are the commands necessary? I really do not know the metadb and am wondering if I need to break all mirrored file systems ( 4 of them) before attempting to recover the var file system. Of course vi, writes to /var/tmp - so another question if vfstab needs to be modified is there another tool?

 

[bfitzmai]

Try not using your meta devices with fsck. Use the raw disk devices:

/dev/rdsk/c#t#d#s#

Try doing the fsck on the raw disk devicd for d6 first and let d7 sync up to it.

 

[GallopingGhost]

bfitzmai,
Thanks for the info - but it was a no go. I took a var filesystem off of a SunBlade and am using that to get the server back into operation. Not a very stable approach, but I may have lost the original var fs. Looks like it may be an OS reload unless I can figure out the metadevice stuff.

 

[daFranze]

if fsck /dev/md/rdsk/dx does not fix the superblock errors the FS is heavily damaged, it would not help to run fsck on the physical devices, they have the same errors, too.

I suggest to restore the /var FS from Tape. if you can restore this with OS Commands (dd, ufsrestore, tar) you are fine (boot from CD run newfs on the metadevice), if you need an application (eg Legato) I suggest to restore it to another host, share it via NFS and copy it to the local disk.

Pressing thumbs! ;-)

Best Regards, Franz
--
Solaris System Manager from Munich, Germany
I used to work for Sun Microsystems Support (EMEA) for 5 years