Update op system, apply patches now not reboot!

[Peter Gillis]

HI, ok tis I ...again
I have updated our rp2450 from ux11.0 to ux 11i. That appears to have gone ok. System rebooted ok. Noticed that /opt was 94% full and /usr was at 50%. so I thought probably should enlarge /opt and /usr.I did. Then tried installing GOLDQPK11i off supportplus cd. In the swagent.log, can see that system was supposed to reboot:
ERROR: The kernel build has failed; the system is not rebooting.
The old reboot command has been saved in the location
"/sbin/reboot.prev".
Also, in the swagent.log there are a few obvious errors like: product not compatible with the new op sys; but there are quite a few errors with :
ERROR: File "/opt/IBMdb2/V7.1/bnd/db2tstfp.bnd" should have cksum
"3309865694" but the actual cksum is "4002250480". ; and

ERROR: File "/opt/IBMdb2/V7.1/bnd/db2spcat.bnd" should have mtime
"987549739" but the actual mtime is "1037380377".

Should I be moving the /system/statnd.prev and /system/vmunix.prev to the original names and rebooting the system from them? How can I work out what has stopped the reboot in the first place? Or, is it because I have enlarged /usr and /opt filesystes, between update-ux and applying goldqpk11i. Is this also the reason why I have got these error messages with mtime and chksum references?

I know there is alot here, but I really, really would appreciate your assistance?

Regards
Maria.

[Bharat Katkar]

Hi Maria,
Suggest you to backup your Data properly, possibly get Ignite backup of VG00 as well and simply proceed with cold install.
That's all
Regards,

[Herbert Loevenich]

Hello Maria,

a backup as in the previous response is always a good roll back procedure.

Additional have a look at the size of /stand
during the swinstall procedure you will have
three kernels in this directory.
And you might run out of space in /stand

See how many /stand/vmunix.*
you have have. Move one out to /
to be sure you can still boot from it.

Hope this helps

Herbert

[Cesare Salvioni]

Hi Maria

seems to me that your main problem has nothing to do with the error in swagent.log you reported.
I mean that kernel build failed is the main problem you have to solve and look for more info about this problem.
One reason for this could be, as already suggested, not enough space in /stand. I also suggest to check how many vmunix files you have there, the only one necessary is vmunix itself but in anormal system you have at least a copy vmunix.prev to be able to recover an unbootable system. By the way DO NOT MOVE vmunix from /stand. It's impossible to boot directly from / if it is a different file system from /stand.
DO NOT DELETE anything in /stand if you don't know exactly what u r doing: wrong deletion leads exactly to your problem: impossible to build kernel
To extend /stand only way is to make a tape recovery (make_tape_recovery), boot from it, stop the boot process and change the size of /stand. This action is like reinstalling the system keeping the actual configuration. Obviously i would make a good backup of all important data before.

Another reason, the worst case, could be some library or object system file missing: this would be really bad news !!!

[Michael Duthie]

Maria,

You need more info on why the kernel build failed. Exit to a shell.

cd /stand/build
rm -r *
cp ../stand .
mk_kernel -s system

Hopefully you should see what the kernel build error is.

[Peter Gillis]

Michael, sorry, but I am pretty new to the kernel process functions. And I cant see why you are running lines 2 & 3 as you suggested.
cd /stand/build
rm -r *
cp ../stand .
mk_kernel -s system

Would you please say why suggest it?

Thanks
Maria

[Peter Gillis]

Bharat,
As far as I know I have backed up my data properly. The actual update to 11.i has gone through. The reboot is not occuring after the install of the required patches.
This is a summary of steps taken:

1.full system backup - Fbackup
2.create mk_recovery tape
3.firmware upg.
4.update-ux
5.swverify
6.full system backup - fbackup
7.create mk_recovery tape
8.increase /opt & /usr to 4 gb
9.Install GOLDQPK11i - at end of install supposed to reboot system auto - this does not occur.

Should I be running:
mk_kernel -s /stand/system
kmupdate
shutdown -r

or will that stuff things up even further.

Thasnks

Maria

[Timothy Chen_1]

Hi, Maria

I just upgrade a R390 server from 11 to 11i using update-ux. After the update-ux, I have increase the size of /opt before applying GOLD patch bundle, same with you, right? But my upgrade and patch installation are fine.

It's seems that your problem is the kernel making issue. Just try to make your new kernel manually following command and to solve this problem first.
# cd /stand
# mk_kernel -s /stand/system
# kmupdate
# shutdown -ry 0

[zhi wang_1]

I am having a similar problem: I installed the latest HWEnable11i_B.11.11.0412.5 on my rp7420. Before I did that my system was running fine and I was able to reboot it without any problem. As soon as I installed the patch, the system will hang in the >ISL prompt. I have to log in through the console and mannually reboot it. The swagent.log does not have any error. Setboot tells me the autoboot is on and autosearch is on as well.

Any idea what happened?

Thanks

Zhi

[Timothy Chen_1]

Hi, zhi wang

It's likely that your server lost the autoboot information. Just follow this example to check it:

# cat /stand/bootconf
l /dev/dsk/c0t6d0
l /dev/dsk/c1t6d0
# /usr/bin/lifcp /dev/dsk/c0t6d0:AUTO -
hpux -lq (;0)/stand/vmunix
# lifcp /dev/dsk/c1t6d0:AUTO -
hpux -lq (;0)/stand/vmunix

If you cannot see the "hpux ..." string, you miss your boot information.

If it is true, just use "mkboot" command to re-write it:
# /usr/sbin/mkboot -a "hpux (;0)/stand/vmunix" /dev/rdsk/c0t6d0
# /usr/sbin/mkboot -a "hpux (;0)/stand/vmunix" /dev/rdsk/c1t6d0

Hope it would help you.