HPUX11 - Boot Problem

[Gavin Davies]

Hi,

I know this problem has come up before on this forum, but I have tried everything that everyone suggested and still haven't had any luck. When I try to boot up my system, it gives me the msg: "Exec failed: Cannot find /stand/dmunix or /vmunix" then goes to the ISL> Prompt.

I have done a "hpux sea" at this point but the msg: "disk(2/0/1.6.0.0.0.0.0;0)sea: cannot open or not executable - Exec failed: no such file or directory"

When I do a "hpux ll" the /stand directory is displayed as: "stat failure on disk(2/0/1.6.0.0.0.0.0;0)/./stand+

I have tried booting into maintenance mode (hpux -lm), but get the original message.

Any sort of help would be appreciated. Thanks!

[Bill McNAMARA_1]

You didn't happen to mv the /stand directory to something else.

You need to boot off the support + cd, mount your / disk and then repair the /stand directory.

If you had an ignite tape it would have been easier.

There's no boot mode you can go into without a kernel...

Later,
Bill

[Gavin Davies]

I haven't moved or changed any directories on my system. Basically, it worked one day, I came back to use it the next, and I was faced with this problem.

The only problem with using the support cd to boot off, is that I can't find it. I was hoping that someone had fixed this problem without the use of this. I can get hold of one, but I need to fix this problem quickly.

Cheers

[Bill McNAMARA_1]

At the initial console prompt, not at isl,
ie: before isl,
issue a sea
Main Menu: Command Menu> SEA
to try to boot a mirror.

What is interesting is that you are getting ISL, which lives on disk too.. so there is not a disk failure. The /stand system is possibly corrupted for some reason (panic?) and you may need to fsck/ restore.

The only method you have is to boot alternate support/ignite.

You might want to have a look at your GSP error logs to see if there was a panic or some h/w failure that caused your problem.

good luck,
bill

[Gavin Davies]

My first impressions were that /stand directory was possibly corrupted.

I have tried a sea ipl at the initial console prompt and I am only displayed with one option (which doesn't work when I try booting from it). I have tried issuing a sea, and booting from the other disk that is available, but it complains that it cannot load IPL, error detected (which I expected).

Thanks!

[Peggy Fong]

You mentioned trying many things, did you try booting from a previous kernel

At ISL> hpux (;0)/stand/vmunix.prev

Just a thought.

Peggy

[Wieslaw Krajewski]

Hi,

Support disk was used till HPUX10. Starting from HPUX11 you should use Install Core OS disk.
Boot from this disk, and select no interaction with ISL.
After a while you should see a menu. Select Run a recovery shell. After next while you should see the support main menu and can select from this menu: recover sn unbootable hpux. From the next submenu can select one of a few scenarios. One of them relates to the kernel. Of course this is only minimal kernel to boot next your system and either generate suitable kernel or recover it from backup.
From your info it follows that it seems impossible to boot from backup kernel.

Rgds.

[Gavin Davies]

Yep. Tried booting from the backup kernel, but the /stand directory seems to be corrupted in some way.

Thanks for the info Wieslaw Krajewski. It looks like I will have to fix the problem with the Core OS disk.

Another point to make is that I can't run fsck, because doesn't this requires certain folders to be in place? i.e the directory where the kernel exists.

Cheers

[KapilRaj]

hi ,

my contribution to you,

Do you have a copy of old kernel (it may be there in a root backup-set. Or copy vmunix from another same type of machine on to a DAT.

Boot through Install / Support CD --->run a shell ----->
#mkdir /boot
#mount /dev/dsk/cXtXdXs1LVM /boot
the above command will mount your /stand filesystem to /boot directory. (cXtXdX is device file of your root disk)
#tar xvf /d/ev/rmt/0m . (Restore vmunix from backup)
# reboot


hope this helps,

kaps

[Gavin Davies]

I have another machine, but it is running HPUX10.20. Does this use the same kernel? Is it possible to boot over a LAN? (Anyone know the correct syntax for doing this?)

I haven't got a DAT, so I can't copy the kernel/vmunix over.

Thanks

[Bill McNAMARA_1]

The recovery CD mount a / in memory.
When you mount your real disk root you will be able to fsck -F hfs /i/forget/the/mount/stand
If your root doesnt mount that may be what needs fsck'ing and not /stand.

In either case the recovery CD should do the trick. Once it does, make an ignite backup!!
(with the latest http://software.hp.com/products/IUX/ )

Bill

[KapilRaj]

i do not know how to make a LAN boot but if you install IGNITE -UX on the other machine you can boot through that over lan

kaps

[Gavin Davies]

I am going to get the HPUX 11 media, and try to fix the kernel. Another thing to point out is: When I do ISL > hpux ll, I also get "stat failure on disk(2/0/1.6.0.0.0.0.0;0)/./lib+"

and

"stat failure on disk(2/0/1.6.0.0.0.0.0;0)/./bin+"

Any tips/steps of what I will need to fix this problem would be greatly appreciated. Although the kernel could possibly be fixed, what about the other directories?

Cheers everyone

[Gavin Davies]

I have managed to get my box up and running using the Core OS disk, but...

I can't telnet to the box. (I can ftp it and use X-Windows).

Is this to do with the kernel not being configured properly? If so, does anyone know how to configure it, or know where there are any instructions?

Another point - when you issue a lanscan command, the message - no namelist is displayed. You also get this message when you try to do anything to do with the kernel or network settings using SAM.

Please help...

Thanks,
Gavin

[Bill McNAMARA_1]

That's good news so far,
have you found out why from the OLDsyslog and syslog.log and shutdownlog why the server went down in the first place?

Regarding this problem:
Issue the mount command just to ensure all youre filesystems are up.

If you replaced your kernel with one on the coreOs, you will have patches not compiled into the kernel that used to be there..
I think you may have to reinstall patches...
that used to be there.

You might want to verify too that set_parms works and that you can set the ip address, hostname and so on.
If you did an fsck on root fs, there may be some corrupted files. Have a look at the /fs/lost+found/ directory for any potentially damaged system files.

file /fs/lost+found/1234343file
if not binary, vi it and see if it looks like a damaged system file.

Later,
Bill

[Gavin Davies]

Couldn't find out why the server went down in the first place.

Ran the mount command, and all filesystems are up.

set_parms works, and the information is already correct (ip address, host name etc)

No corrupted files found in the lost+found directory.

This does look like this is a problem with the kernel, but not sure what to do to fix it, and don't know how to reinstall patches. (I never actually built the machine, so I wouldn't know what patches to install anyway).

Help!!!!

Thanks

[Bill McNAMARA_1]

issue a
#what /stand/vmunix

and verify that all PHKL_ s and PHNE_ 's listed in /var/adm/sw/products/
are installed..

What You don't want to see is products in /var/adm/sw/save/ in the kernel.

(can someone verify this)!

Another command to list current installed software is swlist You want to look at the the state of the installed software (swlist -l filesets -a state I believe)

Just wait before someone else verifies that this is indeed the problem.. I have only an idea on what to do in this case, but have never done it myself. The solution may be simply to apply or delete a driver from the kernel...

What exactly did you do to rescue your system: did you fsck or simply replace a kernel.. The same issue exists in replacing a kernel as booting of a .prev kernel, there will be s/w mismatches and differences will have to be dealt with.

Do you have a recent backup of your kernel (/stand/vmunix) on your backup archives somewhere?

Later,
Bill

[Gavin Davies]

issued a
#what /stand/vmunix

and all PHKL_ s and PHNE_ 's listed in /var/adm/sw/products/ seem to be installed.
This generated hundreds of files, so I haven't checked in any detail.

To rescue my system, using the OS core disk, I chose 'Replace only the kernel on the root file system'. To do this, you are forced to run an fsck first.

Don't have a backup of the kernel anywhere.

Cheers

[Bill McNAMARA_1]

Is there a vmunix.prev in /stand

what /stand/vmunix > what1
what /stand/vmunix.prev > what2

diff what1 what2
(obtain PHKL, PHNE that was in what2 and not in what1.
Look at /var/adm/sw/logs/swagent.log
for recently added patches and get them too.
swcopy to create a depot directory, then
swinstall all the patches.

The patches can be downloaded individually from itrc.hp.com home. and swcopy information can be found by using a quick search from here too.

Keep us informed..
Later,
Bill

[Gavin Davies]

Haven't got a vmunix.prev

[Bill McNAMARA_1]

Post up the output of the error from the samlog (var/adm/samlog/samlog.log ) and the current syslog.log..
We should be able to identify from the errors what is needed.

Bill

[Gavin Davies]

There is no samlog directory, but I have attached the current syslog.log file.

Cheers

[Bill McNAMARA_1]

The problem is NFS related possibly because of this error.
Automount abort!!!
May 22 16:49:37 STAHP01 automount[414]: Possible cause:
May 22 16:49:37 STAHP01 automount[414]: 1)No map provided
May 22 16:49:37 STAHP01 automount[414]: 2)Using the same map as the running automount
May 22 16:49:37 STAHP01 automount[414]: 3)The maps contain invalid mount points

That's why when you telnet it fails... it probably fails at login because home is nfs mounted?

I would imagine the source of the problem is here:
]: Cannot establish NFS service over udp: bind problem. Attempting older NFSD binding method.

May 22 16:50:34 STAHP01 vmunix:

WARNING: modpath: DLKM is not initialized

So let's try to track down this error on the itrc..

DLKMs btw are dynamically loadable kernel modules.

Later,
Bill

[Bill McNAMARA_1]

Try following Andy monks suggestion here:

http://forums.itrc.hp.com/cm/QuestionAnswer/1,1150,0xd36637f45ef7d4118fef0090279cd0f9,00.html

It may be a little dangerous, so I reckon you should do a backup before, although there is a chance that the kernel will be updated correctly according to your system and not the recovery cd.

Bill

[Gavin Davies]

Just another point to mention:

In the /stand directory, I only have two files - 'vmunix' and 'ioconfig'. There is also a directory in here (Build) that contains the 'System' file, but the size of this is 0k.

Therefore, when running mk_kernel -s ../system it will probably fail.

Shouldn't there be more files in the stand directory?

Thanks