Lost Data Lost Admin

First time so excuse my ignorance please.
I may not be accurately describing the issue.
I have inherited a small lab mostly SUN V120s.
We lost power and are trying to recover.
Nope no backups...
The primary issue I have is 1 box is an Oracle Server.
It has 2 36Gb harddrives.
I am able to reboot the system but it will not mount a "meta device".
The vfstab file indicates a device mount of /dev/md/dsk/do on /oracle
There are several directories under root (e.g., /U01) that are linked to
/oracle.
Apparently this device according to the /etc/lvm/md.cf file looks like this:
d0 3 1 c1t0d0s4 \
1 c1t1d0s0 \
1 c1t1d0s1

I can see that this 'device' is using partitions from both harddrives
and I think the second harddrive (t1) has been trashed.

I know I am all over the board here, I am fairly new to
Solaris SysAdmin but I figured it couldn't hurt to try this
method for any help I can get.

Much obliged
murphsr

The utility that creates those metadevices is called "Solaris Disk Suite". I know very little about it (almost all boxes I admin use Veritas instead of SDS for disk management), but it is a common tool. I'm sure you could find lots of information about it at docs.sun.com.

For example, here is the guide for the latest version:
http://docs.sun.com/app/docs/doc/805-5962?q=solaris\+disksuite

You may need to check a different doc, but there are tons there so you should be able to find the one for the version of SDS on your system.

Suggest running fsck on the device (/dev/md/dsk/d0) and see if that fixes it. It may not if Oracle was using raw partitions. If not (and you are SURE that it's t1 that is messed up), you may be out of luck since it isn't a mirror - it shows that the /oracle (or /dev/md/dsk/d0) was made up of c1t0d0s4, c1t1d0s0, and c1t1d0s1. If two of those are now damaged (c1t1), you may not be able to fix it.

Was there any other line in md.cf that contained the d0 device? (I'm hoping you had a mirror - something like this - I doubt it since you state there are only 2 36GB drives)
d10 -m d0 d1 1
d0 1 3 c1t0d0s4 c1t1d0s0 c1t1d0s1
d1 1 3 c1t2d0s4 c1t3d0s0 c1t3d0s1

If so, you can break the mirror, recreate the d0 after replacing the hard drive (as long as that was all that was on it).

Since you stated the md.cf was in /etc/lvm, you are running at least Solaris 8 (maybe 7 - I don't recall) since the md config files were moved from /etc/opt/SUNWmd to /etc/lvm around that release. It's always helpful to state the OS and version.

Thank you,
Appreciate the replie(s).
Yes I am runing Solaris 8
I was able to connect with a Unix Admin
working another project and he is looking into it.
He has apparently been able to run the metainit -f-a and recreate the
device. FSCK is running against it now but he doesn't feel the
outlook is very promising.
Nope d0 was not repeated in the md.cf
If there is a silver lining...certainly this forum and
more knowledge is that.

murphsr