UNIX for Advanced & Expert Users

I installed sun recommended cluster patches for solaris 2.5 on my
solaris 2.5 server and then issued a shutdown -i6 -y -g0 to reboot the server but server failed to come back up.

the server tries to boot up and goes through initialising memory then server freeze.

I try to bring the server up by

Pressing the stop -a keys on the keybord, which bring me to the ok prompt

then issued

boot cdrom -s to boot from cdrom

but what do I do from here I have never been in this situation before.

Can you Help please


Thanks in Advance

I think its better to try to boot single user and find out what kind of problem prefent your box from booting.

So at the ok> prompt
boot <bootdisk> -s

if this fails also you can boot from CDROM

I have tried to boot the server in single mode (boot -s ) from ok
prompt.

This displayed the following error message:

Bad trap CPU=0 type=0x31 rp=0x10403648 addr=1024 mnu_fsr=0x0


End trackback


what does the above message means? what is the solution to this above problem

Can you help please.

Thank in Advance

When you applied the recommended patches, were you in single user mode?

If not, then more than likely you have 'hosed' the system. One or more patch did not apply correctly because of the state of the server. It is causing a panic on the server.

If the server is actually up in single user - remove the patch.
OR
You could try booting cdrom -s and then mounting your different partitions under /mnt. You could then try to find the patch directory (where you installed from) and run the backoutpatch script (note you would have to change it to let it know that the real /, /usr, /whateverelse is actually under /mnt/, /mnt/usr, ...
If you don't it will fail for two reasons - one: you can't write to the / partition while booted from cdrom - and two, the cdrom doesn't have the patch on it)

If you can't do that - re-install the OS, apply patch in single user, init 0, boot -s (normally I would boot -s after applying a patch - just in case )

Another possible cause - a bad CPU or memory - but I don't find anything on SunSolve except software issues for that error.

I found the test below at http://docs.sun.com/




BAD TRAP
Cause A bad trap can indicate faulty hardware or a mismatch between hardware and its configuration information. Data loss is possible if the problem occurs other than at boot time.
Action If you recently installed new hardware, verify that the software was correctly configured. Check the kernel traceback displayed on the console to see which device generated the trap. If the configuration files are correct, you will probably have to replace the device.
In some cases, the bad trap message indicates a bad or down-rev CPU.
Technical Notes A hardware processor trap occurred, and the kernel trap handler was unable to restore system state. This is a fatal error that usually precedes a panic, after which the system performs a sync, dump, and reboot. The following conditions can cause a bad trap: a system text or data access fault, a system data alignment error, or certain kinds of user software traps.

sun Engineer was callout and after replacing memory and cpu twice he concluded that the problem is not actually an hardware problem.

he said the problem was patches being installed in run level 3 rather than single user mode.

However, I booted the server from cdrom and mounted on /mmt

mount /dev/dsk/c0t0d0s2 /mnt

I checked the installed patches in /mnt/var/sadm/patch and I
could see the recently installed patches but when I issued

patchrm "patch name"

this responded with patch is not installed or is invalid.

but I could see the patch in /mnt/var/sadm/patch

what am I doing wrough? how can I removed the recent installed
patches.

Your original message showed you were running Solaris 2.5. I don't believe patchrm was in the OS back then. I checked a 2.5.1 system and there is no such command.

If you are running 2.5, then you need to go to the directory where you originally installed the patch from - that sould be /tmp. (Problems sometimes occur with installing from other directories). You should have a 103xxx-xx directory. If you saved the files to allow a backout, then there should be a backout script to run (you don't want to use patchrm since that is not set up to remove all the patches added by the one you installed).

If this doesn't work - you have to reinstall the OS - no way around that.