loop boting process -- impossible boot

[Cesar Gonzalez_1]

Hi. I??ve got a problem booting my HP9000. Last weekend there was an power interruption and now server can??t boot, it??s in a continuos loop:

Soft Boot
ISL Booting hpux
.
.
.
.
Logical volume64 0x3 configured as root
Ita was not possible for the kernel to find a process that caused this crash

Dumpsys() called

WARNING Early Dump not configutred, contact HP suplier
WARNING can??t initializedump system
Dumpsys() fialed returning 4294582372

*****************************
System halting during LVM configuration.
.....
.
.
.
.

is not exactly, but that was the most important msgs, and start again the loop.

Any help?!?

[Paula J Frazer-Campbell]

Hi

Have a look here:-

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x6a337b8d1de3d5118ff40090279cd0f9,00.html

HTH

Paula

[Cesar Gonzalez_1]

Well, I did some steps:

First I booted in single-user, but I cant do anything to LVM, because sytem show me:

#pvdisplay /dev/dsk/c0t5d0
pvdisplay :Warning: couldn??t query physical volume /dev/dsk/c0t5d0"
The specified path does not correspond to physical volume attached to this volume group.
.
.
.

Same answer with vgdisplay.
I checked device??s spelling.

Then I decided to boot withou quorum (could somebody explain me what??s exactly meaning of that!! I saw in forum) --> hpux -lq (;0)/stand/vmunix.

After that I reset server and now it boot, but I get next msg:"Failure in attaching PV(8/16/5.5.0) to the root volume group
error no: 16"

and same result than in single-user mode when I make pvdisplay or vgdisplay.

Sugestions?!?

[Paula J Frazer-Campbell]

Hi Cesar

It looks like your power cut could have blown a disk controller.

Options

1. Power down the server and reseat the cables disks and cards then reboot.

2. Place a support call with HP for a replacement of your disk controller card.

HTH

Paula

[gimini]

Hi,
I think your system is not able to save core files after it paniced due to power fail.two reasons i can think of are:
1. The /var/adm/crash directory does not exist.
In this case you have to create one.
2. The /var file system is not big enough to
hold the core files(vmunix and vmcore.n)
under /var/adm/crash directory.
Check both by booting to single user:
1.Power recycle the system.
2.Interupt booting within 10 second.
3.Boot to ISL>
4.Boot to single user e.g hpux -is
5. At # mount /var.
6.Reboot system After checking the above reasons or any other reasons you might find.
Note:You can remove core files since you know the reason for system crash(power fail).
Regards.

[Cesar Gonzalez_1]

Hi gemini.
Well, I taked a view from /var/adm/crash, and the path that told me exists, but is empty.

Really It must be a problem with LVM because the:
"Failure attaching PV (8/16/5.5.0) to the root volume group" appears in all boots.

How can I discover if my system is ready to recover from a crash?

[Ruediger Noack]

Hi,

looks like your disk with adress 8/16/5.5.0 is down or defect because of hard crash during power interruption. You have to boot your system without quorum (see your LVM dokumentation) - hpux -lq ... from ISL prompt.
If this disk is your boot disk and you don't have mirrored theis disk you have to change your disk and to restore with ignite tape.
If your system boots with hpux -lq ... you can try to activate all VGs and analyze your problem (hardware problem or LVM problem).

Good luck
Ruediger

[John Palmer]

Hi,

It looks as though you've lost disk c0t5d0. From the symptoms, I'd guess that your vg00 has two disks, one mirroring the other.

When you lose one disk in a two disk volume group then the vg can't be activated without overriding the quorum requirement (>50% disks must be available) - see man vgchange for more information.

The use of the -lq argument to hpux forces the boot process to vgchange -a y -q n vg00. As you've found out, this is the only way that you can get the system to boot.

I would check the status of your disks with ioscan -fnC disk and if necessary, get the failed disk replaced. You can then follow standard procedures for recovering a lost mirrored boot disk.

Regards,
John

[gimini]

Hi Gonzalez
I had the impression from your initail message that you had problems with core dump configuration.To check your system's automatic core dump is configured properly;
1.open /etc/rc.log file and look for the line
which specifies the target dump directory
/var/adm/crash.You can change this directory
to a more spacy one on other file system if
/var is small (smaller than physical memory).
2. If you find the
two core files in /var/tmp/crash ,then you will know that you system has crashed and configured properly for automatic core dump.

As for your booting problem,have you tried to execute vgscan command?this command allows for recovering of LVM configuration file.Otherwise the disk in question has gone offline somehow.
Regards.
.

[harry d brown jr]

Cesar,

Do you have a "recovery" tape? If not, when you get this issue resolved, I'd suggest creating a "recovery" tape using "make_tape_recovery" which is part of ignite. And I'd do the same for every HPux machine you have.

live free or die
harry

[gimini]

Hi Gonzalez
I had the impression from your initail message
that your problem was related to auto core dump.Anyway,to check if your system is capable of automatic core dump,just make sure that directory /var/adm/crash is specified in /etc/rc.log bootup script.
And check that /var is bigger than physical memory.If not ,edit /etc/rc.log file and spcify another directory in another spacy file system.
As for LVM boot problem,have you tried to execute the vgscan command? this allows for recovery of LVNconfiguration file.
Regards.

[Ravi_8]

Hi,

Once i was facing similar problem. Actually problem was the cable connecting HDD to system board was loose. After tightning, system booted well.

[Cesar Gonzalez_1]

Thanks to all. John was right. My alternative path (5.5) is broken, and system couldn??t boot because LVM was in both disks, but I couldn??t boot because alternative path was incorectly typed. Booting without quorum (hpux -lq(;0)/stand/vmunix) allow the system to boot.

Now I??ll change broken disk.

Jhon, sorry for the points, now I can see It??s ok but before I was testing. You deserve 9 points.

[Carlos Fernandez Riera]

Try to boot from Support CD or tape.

Run a recovery shell and vgimport your disk;

Run fsck on each filesystem and mount. unmount all and reboot.


BUt... it seem to be hardware for me...

Disk, controler or, almost sure, disk??s backplane.