HP UX crash

[Fehime Tüfekçioğlu]

Hi,
Our HP UX 11i v2 server with Oracle 10g installed has crashed. Server gives system panic, it tries to dump, and reboots itself immediately.Then it reboots itself again and again. So we can not even login.
We used the server only for database issues, so we think the problem may be Oracle.

Any help will be appreciated...

Thanks

[Pete Randall]

Interrupt the boot process at the "10 second" prompt and boot into single user mode with "hpux -is". Then try increasing your run level with "init 2", then "init 3". You might want to temporarily skip the oracle startup to prove whether that is the problem or not.


Pete

[Simon Hargrave]

edit /etc/rc.config.d/oracle and change ORACLE_START to 0, this will stop oracle booting on startup. Whether or not oracle is your problem is a different matter.

check /var/adm/crash for crash dumps to send to oracle.

check /etc/shutdown.log for clues.

look in /var/tombstones/ts99 for clues.

[Franky_1]

Hi,

interrupt the boot process, go into single user mode and then mountall
Edit the /etc/rc.config.d/oracle file and set
ORACLE_START=0
Then type init 3 to boot and see if it runs without Oracle.
If so then you'll have to analyze the Oracle Crash-/Logfiles

Regards

Franky

[Fehime Tüfekçioğlu]

Hi again,

I can not get into single mode, because after writing "hpux -is" the same problem occurs. After dumping, the machine reboots itself.

Any more ideas?

[Simon Hargrave]

if you can't get into single user mode, it definately isn't an oracle issue. it's most likely to be a hardware error of some sort. do you have more details of the PANIC message?

most likely you will need HP to replace some hardware.

[Kent Ostby]

Fehime --

A couple of things you can attempt to do.

Interrupt the boot up and go to the boot menu.

If this is an older server you can type PIM to see if an HPMC occurred.

On a newer box, you can look at the logs from the boot menu and see if any HPMCs show up.

Secondly, you could try to boot off of the Install/Support media and select Run A REcovery Shell.

If this crashes then its definitely HW since the Support Media CD is a known good Read Only CD.

Best regards,

Kent M. Ostby

[Olivier Decorse]

Can you try "hpux -lm". May be it is a LVM quorum problem.
Can you post the exact error message, with "blocking" the scroll console ?

Olivier.

[Gerhard Roets]

Hi Fehime


Look on the GSP/MP of the machine when you reboot it. There should be some error messages. If its a software alert you will have to do some more intensehunting, but even with that log a call with HP and they can look at that for you. With a software alert there might be more intense hunting required.

WHat you can try .. but DO NOT MODIFY ANYTHING if you do not know or are unsure of doing is ... boot hpux with "hpux -is -lq". This will remove the requirement of vg00 having to have 51% of disc available to boot.

Regards
Gerhard

[Fehime Tüfekçioğlu]

Hi all,

I scrolled the screen and got the error message, it is something like

SYSTEM PANIC: ALL VTSMOUNT ROOTS failed
sync'ing disks
0 buffers not flushed
0 buffers still dirty
........

with "hpux -lm" the same messages are displayed.

Any more ideas?

[Olivier Decorse]

Oh, sorry : i made a mistake.
You have to try "hpux -lq" for the LVM Quorum ...

Olivier.

[Gerhard Roets]

Fehim

Did you try to boot into your previous kernel.

"hpux /stand/vmunix.prev" ?

If this fails ...

I hope you have an up to date ignite tape on hand if this does nto work ... your stand file system could be corrupt or the disc drive could have failed. This will be the point where you need to look at recovery.

Regards
Gerhard

[Massimo Bianchi]

Hi,
with "VTSROOT MOUNT" message it's time to search for the Kernel CD and start a recovery shell.

I suspect a first bad crash, due to ?unknown? that crashed the "/".


Due to this, it's not even capable of booting.


HTH,
Massimo