Server booting

[Walter Goncalves]

Hello.

I have a K580, running HP-UX 11 which does not want to boot.
We had a power failure (yes our UPS did not kick in) and the server has not come up after that.
It goes through the self tests and starts booting from the internal disk but then it displays the following

32 Processor
34 Processor
36 Processor
38 Processor
49 Memory
btlan4 Initializing 10/100Base-TX card logical volume 64, 0x3 configured as root

[Massimo Bianchi]

Hi,
it is doing the fsck.
It can take may many time, up till 1 hour, in case of large configration of lvol.
The message you saw is normal.

I suggest you to wait.

Massimo

[Enrico P.]

Hi,
what is the code in the display?

Enrico.

[Michael Steele_2]

Need the FLT ##### from the LED panel.

On the same panel are CPU status indicators: CPU 0123 for all 4 up, CPU 0 23 for #1 down, etc.

To check the PDC log turn the key to service, control b, sl. This will display the last 50 LED messages.

[T. M. Louah]

The best approach is to interrupt boot process.
& boot to single user mode
in K class you can turn the server key to SERVICE mode then press CTRL+B on keyboard it will take you to CM prompt.
Now type in rs to reset the machine watch for "interrpt boot ... " press ESC it will take you to the boot handler menu:
- type in "bo pri"
- answer yes to interract with IPL
- at "ISL>" type in "hpux -is" (to boot to maintenance mode type "hpux -lm")
this should take system to single user, only / & /stand are mounted at this point manually mount /usr & /var if it says filesystem corrupt need checking type in "fsck -y /dev/vg00/rlvolX"
Now they should mount (if not you are out of luck) .. then investigate more .

[Walter Goncalves]

The server has been in this state for over 4 hours.
I have tried to boot into single user mode but get the same message/error.

The current code on the LCD display is
INIT CEF4
LAUNCH 0123

[Caesar_3]

Hello!

So in the moment after you write it stop
or continue do something.
What msg you get on LCD on station.

Caesar

[Massimo Bianchi]

Hi, you said it is a K model.

O.K., 4 hours is too much to wait.

Check your key: is it turned in maintenance mode ?

It it is turned so (the little key on the right), it will remain in that state forever, waitin for comeone in console mode to press the key.

I used that trick to have not to wait for the"10 secs" prompt.

You turn the key, and then go to GSP. until you get to console mode again, it won't boot. I didn't remember the messages, thought.

HTH,
Massimo

[Kelli Ward]

Hi,
It seems like hardware but, for grins, you can give booting the alternate kernel a shot.

Break the initial boot sequence.
Boot your primary drive to ISL.
AT ISL type:

ISL> hpux ll /stand

look for another kernel and try booting single user to it.

GL,
Kel

[Elena Leontieva]

Code
Description

CEE0 Kernel was loaded and initialization has begun
CEF0 Kernel has entered main
CEF2 Kernel is about to configure the I/O system
CEF4 Kernel is about to mount the root file system
CEF6 Kernel is about to set up the page-out demon
CEF8 Kernel is about to start the initialization process

[Michael Steele_2]

control b
rs (* reset *)

This will reboot the server from the service processor.

Watch the LED panel during reboot until the FIRST 'FLT' message is displayed. Maybe 5 min. will pass.

Also, from the main menu or BCH, :

sea (* list all devices - 5-10 min. *)
sea ipl (* list boot devices - 5-10 min. *)

bo pri (* boot off of primary disk *)
interact with ipl? (* answer yes *)
ISL> hpux -is (* boot to single user mode *)

See if you can get to single user.

-or-

ISL> hpux -lm -is (* boot to LVM mode *)

If you can't single user, test for bootlif corruption with LVM maintenance mode. If you get here, stop.

From LVM maint. mode you can only reboot or shutdown. Don't mount anything.

[Helen French]

I would be starting from boot console menu in order to troubleshoot:

1) Interrupt boot process and enter to boot console menu.
2) Check hardware information and see if you got any h/w (HPMC) errors
3) Run tests on memory, check the information and see if it detects all modules.
4) Check if all CPU is enabled and working properly
5) Run "search" command to see if it detects all Hard disks including the root disk
6) If needed, run Offline diag on system and make sure there is no problem with the hardware.
7) Now check s/w or OS related issues. Try booting from single user mode or LVM maintenance mode.
8) Fix errors from root prompt. Check file systems, network etc.

[Walter Goncalves]

The key was in the service position. However looking at all the other K's in the computer room this seems prety normal.
However I have turned the key to the ON position and reboot. Will se what happens.
I have tried booting from the alternate disk but it did not help - I got the same error.

What I will try though is to boot off the old kernel.

[Helen French]

From your last post:

The key should be in ON position. You need to put it on the service only when required (maintenance time etc). When the server comes back, try booting from your original kernel and post the error if any.

[Michael Steele_2 · Accepted solution]

WHAT IS THE FIRST 'FLT' MESSAGE ON THE LED!!!!
WHAT IS THE FIRST 'FLT' MESSAGE ON THE LED!!!!
WHAT IS THE FIRST 'FLT' MESSAGE ON THE LED!!!!
WHAT IS THE FIRST 'FLT' MESSAGE ON THE LED!!!!
WHAT IS THE FIRST 'FLT' MESSAGE ON THE LED!!!!

[Walter Goncalves]

There is no FLT message on the LCD.

How do I see what commands I can run on the LVM maintenance prompt ?

[Massimo Bianchi]

Hi,
in maintenance you have the commands in /sbin commands.

first: use "tsm", to have multiple windows
second: try to do a vgchange -a y vg00

if 2 succedd,
try a

"mount -a"


If the lvols get mounted you are O.K., otherwise you can see which lvol hangs... obviously from another window.

HTH,
Massimo

[Helen French]

Most of the LVM related commands will work on LVM maintenance mode. Try doing a "mount -a" once you get the prompt. If it mounts the root FSs, then you will have almost all diag commands it hand. Check boot defenitions (lvlnboot -v), vgdisplays and lvdisplays. Fix file system errors with fsck.

[Walter Goncalves]

Thanks gents

It turned out to be a faulty root disk. I found this out by going into LVM maintenance and then doing fsck's, mounts etc.

The root disk had SWAP on it I think that is why the server did no want to boot.

[Michael Steele_2]

I'm glad that your able to boot.

FLT messages: As you boot up and also observe the LED panel at the same time you will see TENS of messages scrolling past. Too many to track. You'll see INIT, TEST, RUN, FLT, WARN, etc. These messages change every second. That's why the last 50 are stored in your PDC log under 'sl', show logs.

When a FLT is displayed, AND There's ALWAYS A FLT, when the K class server is unbootable, it will remain displayed on the LCD long enough for you to record it.

LVM maintenance mode. Never mount from LVM maintenance mode. If you'd have read my message above in its entirety, you'd have seen this comment. For LVM maintenance mode only uses the /stand/rootconf file to boot up and bypasses bootlif record.

And corruption is likely, but not always the case, when you mount your file systems or init 1,2 or 3.

So do a sanity check of your system and reboot from primary and alternate and run 'check_patches' or 'swverify' and verify the O/S.