topic Re: Boot off mirror in Operating System - HP-UX

Boot off mirror

Tom Danzig — Sat, 19 Aug 2000 19:24:23 GMT

I can't boot of the mirror disk on V2250 running HPUX 11.00. When I try "boot alt" after interrupting the boot process, I get

"Can't find default file BOOT_TC".

Strange.

I recreated the mirror. After reducing the root vg, I did a dd to overwrite the LIF area on the mirror disk (I did get and IO error when I did this) and then did the pvcreate -f -B cXtXdX, mkboot, etc.

lifls of AUTO on the mirror shows the correct default file

(hpux -lq (;0)/stand/vmunix). Also tried "hpux -lq"

LIF volume headed is OK (LIF10)

Becauce of the IO error on the dd to the mirror and the weird error on alt disk boot attempt, I believe the disk may have bad sectors in the LIF area.

Thoughts / comments?

Thanks,
Tom

Re: Boot off mirror

Antoanetta Naghiu — Sun, 20 Aug 2000 00:54:00 GMT

To have a bootable disk you need to have mirrored /, swap and /stand
1. Do you use LVM and mirroring? If yes, run lvlnboot -v.
2.Take a look on the latest LVM patches and hardware/ critical patches for V class machines. Install Diagnostics and run STM.

Re: Boot off mirror

Stefan Farrelly — Sun, 20 Aug 2000 12:32:13 GMT

unusual that you would use dd to overwrite the lif area, and more unusual to get an i/o error. To check the mirror disk to see if its got any bad blocks etc. do;
dd if=/dev/rdsk/cxxxxx of=/dev/null bs=1024k

If this command works without any errors then the disk looks fine, if not the disk may be bad, try to mediainit first then repeat the dd.

When you create the mirror make sure you do it the correct way, pvcreate -B first, then mkboot then lvextend all of vg00's lvols in order, 1,2,3... then lvlnboot -v should show both disks as bootable and swap/root/dump listed on both disks, then it should boot absolutely fine off the mirror - as long as the primary disk is also present.

Re: Boot off mirror

Tom Danzig — Mon, 21 Aug 2000 11:45:13 GMT

The following outlines the commands/steps used:
lvreduce -m 0 /dev/vg00/lvol1 /dev/dsk/c8t6d0
lvreduce -m 0 /dev/vg00/lvol2 /dev/dsk/c8t6d0
lvreduce -m 0 /dev/vg00/lvol3 /dev/dsk/c8t6d0
lvreduce -m 0 /dev/vg00/lvol4 /dev/dsk/c8t6d0
lvreduce -m 0 /dev/vg00/lvol5 /dev/dsk/c8t6d0
lvreduce -m 0 /dev/vg00/lvol6 /dev/dsk/c8t6d0
lvreduce -m 0 /dev/vg00/lvol7 /dev/dsk/c8t6d0
lvreduce -m 0 /dev/vg00/lvol8 /dev/dsk/c8t6d0
vgreduce /dev/vg00 /dev/dsk/c8t6d0
dd if=/stand/vmunix of=/dev/rdsk/c8t6d0 bs=4096
pvcreate -f -B /dev/rdsk/c8t6d0
vgextend /dev/vg00 /dev/dsk/c8t6d0
mkboot /dev/rdsk/c8t6d0
mkboot -a "hpux -lq (;0)/stand/vmunix" /dev/rdsk/c8t6d0
lvextend -m 1 /dev/vg00/lvol1 /dev/dsk/c8t6d0
lvextend -m 1 /dev/vg00/lvol2 /dev/dsk/c8t6d0
lvextend -m 1 /dev/vg00/lvol3 /dev/dsk/c8t6d0
lvextend -m 1 /dev/vg00/lvol4 /dev/dsk/c8t6d0
lvextend -m 1 /dev/vg00/lvol5 /dev/dsk/c8t6d0
lvextend -m 1 /dev/vg00/lvol6 /dev/dsk/c8t6d0
lvextend -m 1 /dev/vg00/lvol7 /dev/dsk/c8t6d0
lvextend -m 1 /dev/vg00/lvol8 /dev/dsk/c8t6d0
lvlnboot -r /dev/vg00/lvol3
lvlnboot -s /dev/vg00/lvol2
lvlnboot -d /dev/vg00/lvol2
lvlnboot -b /dev/vg00/lvol1
lvlnboot -R /dev/vg00
lvlnboot -v
setboot

The dd to the mirror was used to make sure that the LIF was recreated; not simply checked/updated. This recommendation came from HP.

Re: Boot off mirror

Stefan Farrelly — Mon, 21 Aug 2000 11:50:50 GMT

The commands you listed look ok. Did you try doing a dd on the mirror as suggested? did it work or did you get an error ?

Re: Boot off mirror

John Palmer — Mon, 21 Aug 2000 12:07:07 GMT

I agree that the commands listed look ok.

You could also have installed the diagnostic LIF files with the command:-

mkboot -b /usr/sbin/diag/lif/updatediaglif -p ISL -p AUTO -p HPUX -p LABEL /dev/rdsk/c8t6d0

Are you certain that the 'setboot' path for alternate boot is correct? What is the output from 'setboot'?

Regards,
John

Re: Boot off mirror

Antoanetta Naghiu — Mon, 21 Aug 2000 12:14:56 GMT

Why dd? You do not need it. Repeat the steps, without dd. Eventually, by the end, vgchange -a y /dev/vg00.

Re: Boot off mirror

Patrick Wessel — Mon, 21 Aug 2000 12:28:22 GMT

Tom,
Let's take a look on the problem from a different angel.
Is f-code enabled for the slot with your mirrored boot-path? And are the boot-file and argument set right?
These are causes why I saw the boot_tc message

Re: Boot off mirror

Tom Danzig — Mon, 21 Aug 2000 12:57:21 GMT

# dd if=/dev/rdsk/c8t6d0 of=/dev/null bs=1024
4194157+0 records in
4194157+0 records out

# lvlnboot -v
Boot Definitions for Volume Group /dev/vg00:
Physical Volumes belonging in Root Volume Group:
/dev/dsk/c1t6d0 (1/1/0.6.0) -- Boot Disk
/dev/dsk/c8t6d0 (6/2/0.6.0) -- Boot Disk
Boot: lvol1 on: /dev/dsk/c1t6d0
/dev/dsk/c8t6d0
Root: lvol3 on: /dev/dsk/c1t6d0
/dev/dsk/c8t6d0
Swap: lvol2 on: /dev/dsk/c1t6d0
/dev/dsk/c8t6d0
Dump: lvol2 on: /dev/dsk/c1t6d0, 0

# setboot
Primary bootpath : 1/1/0.6.0
Alternate bootpath : 6/2/0.6.0

Autoboot is ON (enabled)
Autosearch is ON (enabled)

Re: Boot off mirror

Tom Danzig — Mon, 21 Aug 2000 13:18:42 GMT

Patrick,

Primary boot disk:

# lifcp /dev/rdsk/c1t6d0:AUTO -
hpux

Alternate boot disk:

# lifcp /dev/rdsk/c8t6d0:AUTO -
hpux -lq

After first trying "hpux -lq (;0)/stand/vmunix" I then tried "hpux -lq" which is the current AUTO file contents on the alternate disk.

Im not sure all PCI slots are fcode enabled. Can I check this with the system up or only at the boot prompt? Since I know that this machine can boot of of SCSI tape, I suspect that all PCI slots are fcode enabled.

Thanks,
Tom

Re: Boot off mirror

Patrick Wessel — Mon, 21 Aug 2000 13:31:54 GMT

Tom,
Unfortunately is the only reliable way to verify the fcode-enable by doing a printenv at OBP.
Since Teststation Software 5.2 prints POST a message for every slot which is not fcode-enabled. You can see a message like:
PCI 4 SLOT 1 is not in fcode-mask; slot not probed.
In this example is 4/1 not fcode enabled.

You can search in the /spp/data/consolelog for this message. At least the slots with Fibre Channel interfaces should appear

Can you please post this part of the boot-up?
Device : /pci@fe,10000/symbios@0,0/sd@6,0:cntl
File : hpux
Arguments : hpux
Loading : hpux ........... 163808 bytes loaded.
102368 + 61440 + 864504 start 0xd01cc0

I assume the error occurs after Loading:? right?

Re: Boot off mirror

Tom Danzig — Mon, 21 Aug 2000 13:58:23 GMT

Patrick,

Command: boot alt
Device : 6/2/0.6.0
File : BOOT_TC
Arguments : -lq (;0)/stand/vmunix
Loading : BOOT_TC
Cannot open default file BOOT_TC
Aborting loadoperation.

Command: boot pri
Device : 1/1/0.6.0
File : hpux
Arguments : hpux
Loading : hpux ........... 171000 bytes loaded.
105464 + 65536 + 863128 start 0xd023f0

Boot
: disc(1/1/0.6.0;0)/stand/vmunix
7024640 + 834608 + 2357216 start 0x23f168

--------------------------------
Also saw this one once as well:

Command: boot alt
Device : 6/2/0.6.0
File : BOOT_TC
Arguments : LIF -lq
Loading : BOOT_TC
Cannot open default file BOOT_TC
Aborting load operation.

Thanks for your time.

Re: Boot off mirror

Patrick Wessel — Mon, 21 Aug 2000 14:21:26 GMT

Tom,

That looks strange. I have no idea why the boot-file is boot_tc. I would check it in OBP:
printenv boot-file
printenv alternate-file

I'm almost sure that the alternate file is set wrong, because the system boots up from primary without any problems

To set it right it's simple:
setenv alternate-file hpux

Re: Boot off mirror

Tom Danzig — Mon, 21 Aug 2000 14:30:12 GMT

Thanks Patrick. I'll give your suggestion a try at the earliest oppertunity.