Ultra 5 Solaris 9 Install Woes

Question

OK - Like a lot of other lost souls here, I bought my hardware on eBay 'cuz it was cheap & I wanted to try some of this legendary Sun hardware. I've been rather disappointed so far, but not ready to give up yet. Here's my history: Successfully installed & ran Solaris 8 for a couple of months last year, then upgraded to Solaris 9 when it came out. Installations & upgrades were awkward & had to be repeated to get them right, but until about 2 weeks ago I had a Solaris 9 system that seemed pretty stable.
Then something happened - don't have any idea what. The system suddenly started bouncing incoming email... I could access it remotely using SSH, but not from the console. So - I figgered it just needed a re-boot.
Tried this several times with no luck. The boot process would halt after a variety of error mesasges; one was words to the effect that some file wasn't a valid ELF file, was not executable, etc. I thought I must have corrupted the kernel somehow. Oh what the heck I thought, I'll just re-install from cd.
That was 3 days ago. Since then I've been unable to get the system back up after numerous failed install attempts. Each time I run fsck it finds several errors that need correcting. I correct, re-try, but still no luck. I'm beginning to think that I've got toasted hardware (hard drive?), or that there's some corrupted files on the drive that the installer is not overwriting (Yes, I am doing a 'fresh' install each time).
Now I've learned that the IDE drive in this system is not really supported by Sun (can't format the drive in-place). I tried running some diagnostics from the format utility, but it won't allow some of these as I can't seem to umount all partitions (tried umountall, but still cannot unmount all partitions). I was able to get the analyze->read to run - it reported no problems with the hd.
I've tried a little self-help reading, too. Bought a copy of Paul Watters' book "Solaris 9 Admin, A Beginner's Guide"... it hasn't been much help & appears to suffer inaccuracies & incompleteness. So, I'm stalled out - thought I'd try posting here & requesting some advice & counsel. Is there a place for cheapskates like me to go to get help with their raggedy-assed Ultra 5's?
Thanks,
Jay Moore

Answer 1

This system is starting to annoy me also :-) (by the
way, when you newfs'ed slice 2, this is meant to
represent the whole disk and should never be touched.
This is not a good thing to doodle with :-( You arn't
trying to put a mounted partition on this slice are
you?) I'm sorry - when you said c0t0d0s* I thought you meant all slices. I reckoned the important stuff was outside of the 'overarching' slice s2. Did I blow it?
Before we get into the rest of it, I got the following on the initial boot attempt following power-up:
"Not a valid ELF file.
The file just loaded does not appear to executable"
<<Groan>>
Do a prtvtoc on slice 2 and make sure that the
"First Sector" is 0 and the "Last Sector" is the
greatest number in the list, well there will be two of
them hopefully but nothing bigger than slice two. If
this is ok then;I saved the results of prtvtoc to a floppy... here 'tis in its entirety:
* /dev/dsk/c0t0d0s2 partition map
* Dimensions:
* 512 bytes/sector
* 63 sectors/track
* 16 tracks/cylinder
* 1008 sectors/cylinder
* 16708 cylinders
* 16706 accessible cylinders
* Flags:
* 1: unmountable
* 10: read-only
* Unallocated space:
*     First Sector Last
*     Sector Count Sector
* 16837632 2016 16839647
* First Sector Last
* Partition Tag Flags Sector Count Sector Mount Directory
0 2 00 1049328 5548032 6597359
1 3 01 0 1049328 1049327
2 5 00 0 16839648 16839647
7 8 00 6597360 10240272 16837631
This looked reasonable to me, so I continued:
Probably the best plan I can get from here, assuming
that cd#2 was installed correctly and the system was
booting for the first time (I think it does this from
disk after #2) is;
1) ok> boot cdrom -s
2) fsck -y /dev/rdsk/c0t0d0s0 (hope this is the root
slice!).Received msg: "FREE BLK COUNT(S) WRONG IN SUPERBLK"
A "salvage" was reported, and we moved along...
3) mount /dev/dsk/c0t0d0s0 /a
4) cd /a/etc (watch the a, it is important here) << I did... >>
5) mv /a/etc/path_to_inst /a/etc/path_to_inst.old
6) init 0
7) ok> boot -a (hit return for all questions except
the rebuild path_to_inst one, answer yes for that). Oh
and by the way, if you look at the "Multiple QFE Quad
Cards" in the installation board then Reynald_sw was
told the same thing from someone too :-) This may
answer any problems that you may have on the change
and goes in a bit more depth.
8) You should now have a Solaris login prompt, but if
it goes t*its up again let me know where and we will
see what can be done.No - I'm sorry, but I'm still hosed. I got to the point in the "boot -a" process where it asked for the physical name of the root device[blah, blah]. WHen I hit 'Return' I quickly got about 2-3 screens full of warnings and errors about undefined symbols, and "Cannot load module tcp", etc, etc.
Once again - I appreciate your patience. I'll understand if you want to throw in the towel... I just I hope I haven't dug a deeper hole for myself. I'm learning a lot if that helps motivate you :)
Thanks,
Jay