FLT B800 issue

[MrViking]

Running hpux 10.20 on hp9000.
receiving message on boot of init died with return value 256-system panic init died.

tries to dump memory-does so successfully, then reboots again, then keeps doing same cycle.

Is problem hard drive? could it be the removal of patch files? (done about 1 week to clear space) or just yesterday, we killed an init process that appeared to be running from last week (prior to our weekly scheduled reboot)?

do i have to try & reboot from ignite?

[Michael Steele_2]

Get on the system console and type in these commands:

control b
sl (* show logs *)
e (* error logs *)

Note the time stamp and alert levels and paste in anything from this event.

More than likely you have a bad processor and need to place a HW call.

[MrViking]

can't try suggested console commands, as i can't get to login prompt...system doesn't boot

[Torsten.]

It's not clear to me what you did.

What is the meaning of "removal of patch files"? What exactly did you remove?

What is the meaning of "we killed an init process"?

How many init processes did you see???

CTRL B does probably not work, because the system is an old model and does not have a GSP (what model do you have?).

Bxxx chassis codes are always related to OS panics - did you try to boot into single user mode?

[MrViking]

i removed physical patch files from directory /var/sw on 2/13.

yesterday, running command ps -ef, i saw 3 init processes running, one from the 18th, the 11th and the 4th.
i ran command from root kill -9 pid of the init command from the 4th.
keep in mind our system reboots each monday morning...thus the 4th, 11th and 18th entries...

i'm able to boot using hpux -lm now, and then tried activating vg00 using vgchange, but i get error 'can't attach to volumn /dev/dsk/c0t6d0, cross-device link. vgchange can't query phys vol /dev/dsk/c0t6d0.specified path doesn't correspond to phys vol attached to this vol group. can't activate vg00."

is this do to a bad disk? or can't it find vg00?

[Michael Steele_2]

Hi MrViking:

Well your doing a lot of things wrong. Regarding: "...i removed physical patch files from directory /var/sw on 2/13....".

Did you use the 'rm' command or the 'swremove' command? The correct command is swremove. That's because there are referenced files in other directories to remove. So they all interact together as a unit and must be installed and removed as a unit with 'swinstall' and 'swremove' commands.

Was /var filling up? You did a 'du -k' and saw that /var/sw was a big directory? It's always like this.

Regarding: "...yesterday, running command ps -ef, i saw 3 init processes running, one from the 18th, the 11th and the 4th...".

Make sure you attach your /etc/inittab file. It will need to be examined.



i ran command from root kill -9 pid of the init command from the 4th.
keep in mind our system reboots each monday morning...thus the 4th, 11th and 18th entries...

Regarding: "..i'm able to boot using hpux -lm now,...".

'hpux -lm' is for maintenance mode and can corrupt your boot file systems if you do an init 3 from run level one. If you've done this before then this might be the source of your problem. For your errors appear to indicate a corrupt O/S. So instead, use 'hpux -is -lq' from the ISL prompt. Boot w/out quorum into single user mode. If you have a bad disk then booting w/out quorum will bypass this bad disk.

Once your booted w/out quorum then this is a strong confimation that you've got a one of two bad boot disks. Mount /usr and inspect /etc/inittab, /etc/lvmtab (* use strings *) try an init 2 and then an init 3.

[MrViking]

Michael
thanks for help.
i'm sure i've done a few things awkwardly.
hpux -is -lq does not work for booting.
goes through same process with same error messages.
if i boot using hpux -lm, i can activate other vol groups (vg01, vg02, etc), just not vg00.

i cant mount /usr (says on vg00, which is not active) so i can't look at vg00.

should i still try & reboot from an ignite tape if one's found? am i stuck reinstalling? anything else i can do to check disk to see if needs replacing? fdsk?

thanks.

andrew

[Michael Steele_2]

hpux -lq