Kernel changes using SAM

[Eric Antunes]

And a full bdf (for /, /var, etc...) output...

[Ian Warner]

sh /stand/build=>kmtune -lq st_san_safe
Parameter: st_san_safe
Current: 0
Planned: 0
Default: 0
Minimum: -
Module: -
Version: -
Dynamic: No
sh /stand/build=>kmtune -lq bcvmap_size_factor
Parameter: bcvmap_size_factor
Current: 2
Planned: 2
Default: 2
Minimum: -
Module: -
Version: -
Dynamic: No
sh /stand/build=>

Ian

[Ian Warner]

##@(#) config.sys $Date: 2003/06/13 06:23:05 $Revision: r11.11/7 PATCH_11.11 (PH
KL_29243)



sh /stand/build=>bdf
Filesystem kbytes used avail %used Mounted on
/dev/vg00/lvol3 204800 170272 34408 83% /
/dev/vg00/lvol1 295024 121384 144136 46% /stand
/dev/vg00/lvol8 2097152 1672816 424336 80% /var
/dev/vg00/lvol7 3145728 1470656 1662032 47% /usr
/dev/vg00/lvol4 3145728 1645960 1488080 53% /tmp
/dev/vg00/lvol6 5242880 3617184 1613056 69% /opt
/dev/vg00/lvol5 512000 293576 217104 57% /home
/dev/vg00/emacs 7168000 3169855 3748325 46% /emacs
/dev/vg00/backup 7168000 5048 6717443 0% /backup
sh /stand/build=>


Ian

[Eric Antunes]

What values for flocks (kmtune -lq flocks) and other(s) changed parameters?

Did same returned a error message when changing one of the parameters?

[Eric Antunes]

Ian,

I've no much time today to help you as I would like...

But I think you should rollback with all changes on parameters and, after that, change just one parameter and process kernel: this will give the parameter that is not being accepted...

[Ian Warner]

The only parameter I am trying to change is nflocks from 2048 to 8192.

The results of kmtune is as follows.

sh /stand/build=>kmtune -lq nflocks
Parameter: nflocks
Current: 2048
Planned: 2048
Default: 200
Minimum: -
Module: -
Version: -
Dynamic: No
sh /stand/build=>

Ian

[Stephen Keane]

It would eliminate the 'incorrect setting of a parameter' hypothesis if you revert back to the previous settings and compile the kernel. If the setting of the parameter is the problem, your kernel will build OK. If, however, the kernel still doesn't build either

a) You've not changed the parameter back to how it was.

or

b) You've changed something else as well.

or

c) It's nothing to do with the parameter you are trying to change.

[Eric Antunes]

Hi,

This nflocks must be related with other parameters so you may have to change those too before processing the kernel...

Do a search in ITRC for "nflocks" and you'll get more than 50 threads like those:

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=665998

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=224927

...

[Florian Heigl (new acc)]

Ian - if the stape driver contains the st_san_safe option, please try as I suggested and build a kernel without it.

- Then You'll be able to adjust nflocks as You wanted
- You don't have to adjust other values unless You replaced the hp-ux default formulas
- Please consider posting Your /stand/system

Is Your tape SAN- or SCSI-Attached?

I'll try to find out how to remove the dependency for the latter case, or mabye I find some hints when st_san_safe was introduced, maybe the specific patch needs to be reapplied (overwrite)

[Ian Warner]

Day 3 of the problem.

I removed stape from /stand/system and tried to recompile the kernel. The exact same error returned. The copy of new /stand/system is enclosed.

Help!!!!!!!!!

Ian

[Eric Antunes]

Hi Ian,

The first strange thing I see is that your system file is missing sio and stape at the "* Drivers and Subsystems" part:

...
sdisk
sio
stape
stcpmap

I may be related. Have you the tape drive correctly installed?? Can you post the result of the following command:

#ioscan -fnCtape

Best Regards,

Eric Antunes

[Ian Warner]

Eric,

as requested -
sh /stand=>ioscan -fnCtape
Class I H/W Path Driver S/W State H/W Type Description
=====================================================================
tape 0 0/0/1/0.3.0 stape CLAIMED DEVICE HP C5683A
/dev/rmt/0m /dev/rmt/c0t3d0BESTn
/dev/rmt/0mb /dev/rmt/c0t3d0BESTnb
/dev/rmt/0mn /dev/rmt/c0t3d0DDS
/dev/rmt/0mnb /dev/rmt/c0t3d0DDSb
/dev/rmt/c0t3d0BEST /dev/rmt/c0t3d0DDSn
/dev/rmt/c0t3d0BESTb /dev/rmt/c0t3d0DDSnb

I removed stape fro /build/system as per Florian Heigl's request.

Ian

[Eric Antunes]

Ian,

Run this to check the state of your patches:

#/usr/contrib/bin/check_patches

[Eric Antunes]

And, after posting the previous post result, post this one:

#swlist -l fileset -a state | grep -i corrupt

[Ian Warner]

Eric,

OK, is this what you want?

/usr/contrib/bin/check_patches
Obtaining information on installed patches
Checking installed version of /usr/sbin/swconfig
Checking for invalid patches
Checking object module checksums for active patch fileset 697 of 697
Checking patch filesets for active patch 354 of 354
Checking state for patch fileset 1144 of 1144
Checking patch_state for patch fileset 1144 of 1144
Running swverify on all patch filesets, this may take several minutes
RESULT: Problems found, review /tmp/check_patches.report for details.

========================= Swverify on Patch Filesets =========================

ERROR: The swverify(1M) command on all patch filesets failed with
the following error(s):

ERROR: "harlie:/": 6 files had errors during this operation.

Please review the /var/adm/sw/swagent.log file for details.

==============================================================================

which part of the above log do you wish me to examine?

Ian

[Ian Warner]

sh /=>swlist -l fileset -a state | grep -i corrupt
sh /=>

[Eric Antunes]

Ok, you have bad installed patches. Please attach the following files:

/tmp/check_patches.report
/var/adm/sw/swagent.log

Eric

[Ian Warner]

As requested, Eric.

[Eric Antunes]

OK,

You have a patch with "Warnings":

WARNING: Verified with warnings Ignite-UX.BOOT-KERNEL,l=/,r=B.3.8.201

You shouls check the file description in ITRC later to see if it's critical or not for your case.

But this is not the cause of your presente issue. Here is the bad things from swagent.log:

...
ERROR: File "/usr/lib/lp/model/rmodel.asx" should have mode "444" but
the actual mode is "755".
ERROR: Fileset "PHSS_29031.PRT-LP-JPN-RUN,l=/,r=1.0" had file errors.
ERROR: File "/usr/lib/lp/model/ESCP" should have mode "444" but the
actual mode is "755".
ERROR: File "/usr/lib/lp/model/PCL5.asian" should have mode "444" but
the actual mode is "755".
ERROR: File "/usr/lib/lp/model/PCL5.nloo" should have mode "444" but
the actual mode is "755".
ERROR: Fileset "PHSS_29031.PRT-LP-RUN,l=/,r=1.0" had file errors.
ERROR: File "/usr/lib/lp/model/hpc1200at" should have mode "444" but
the actual mode is "755".
ERROR: File "/usr/lib/lp/model/hpc1205at" should have mode "444" but
the actual mode is "755".
ERROR: Fileset "PHSS_29031.PRT-LP-TCH-RUN,l=/,r=1.0" had file errors
....

Try this:

Change the refered /usr/lib/lp/model/ files to the appropriate mode 444.

Then,

#swremove -i

And select those 3 patch to remove and, then reinstall them if they are applicable to your system.

Hope this will result,

Eric

[Ian Warner]

That cleaned the patches -

Obtaining information on installed patches
Checking installed version of /usr/sbin/swconfig
Checking for invalid patches
Checking object module checksums for active patch fileset 697 of 697
Checking patch filesets for active patch 354 of 354
Checking state for patch fileset 1144 of 1144
Checking patch_state for patch fileset 1144 of 1144
Running swverify on all patch filesets, this may take several minutes
RESULT: No problems found, review /tmp/check_patches.report for details.

However......
sh /stand/build=> -l \ config.SAM.mk /stand/build/system.SAM <
Generating module: krm...
Loading the kernel...
ld: Unsatisfied symbol "st_san_safe" in file /usr/conf/lib/libio.a[mt_prop.o]
ld: Unsatisfied symbol "bcvmap_size_factor" in file /usr/conf/lib/libfs-pdk.a[uf
s_mchdep.o]
2 errors.
*** Error exit code 1

Stop.
config: make did an exit(1)

It still errors........

[Stephen Keane]

What does

# strings /usr/conf/lib/libio.a | grep Revision

# strings /usr/conf/lib/libfs-pdk.a | grep Revision

give you?

[Ian Warner]

Stephen

sh /stand/build=>strings /usr/conf/lib/libio.a | grep Revision
@(#) pat_psm.c $Date: 2003/09/30 22:19:01 $Revision: r11.11/10 PATCH_11.11 (PHKL
_29537)
@(#) pa_pdk_stub.c $Date: 2001/07/23 09:42:31 $Revision: r11.11/1 PATCH_11.11 (P
HKL_24730)
@(#) pa_legacy_psm.c $Date: 2001/10/18 22:38:08 $Revision: r11.11/2 PATCH_11.11
(PHKL_25552)
@(#) pa_generic_psm.c $Date: 2002/09/04 08:38:43 $Revision: r11.11/4 PATCH_11.11
(PHKL_27715)
@(#) pa_ccio_psm.c $Date: 2001/10/30 11:07:35 $Revision: r11.11/1 PATCH_11.11 (P
HKL_25602)
@(#) olar_psm_if.c $Date: 2001/03/01 08:48:44 $Revision: r11.11/1 PATCH_11.11 (P
HKL_23505)
@(#) olar_psm.c $Date: 2001/03/01 08:48:44 $Revision: r11.11/1 PATCH_11.11 (PHKL
_23505)
@(#) mt_prop.c $Date: 2002/04/08 15:53:15 $Revision: r11.11/2 PATCH_11.11 (PHKL_
26785)
@(#) ike_psm.c $Date: 2004/03/11 12:53:57 $Revision: r11.11/13 PATCH_11.11 (PHKL
_30067)
@(#) gio_search.c $Date: 2000/12/11 16:38:12 $Revision: r11.11/1 PATCH_11.11 (PH
KL_23196)
@(#) gio_node.c $Date: 2002/12/20 11:47:56 $Revision: r11.11/3 PATCH_11.11 (PHKL
_28209)
@(#) gio_modfuncs.c $Date: 2003/09/28 10:32:54 $Revision: r11.11/2 PATCH_11.11 (
PHKL_29818)
@(#) gio_kdev.c $Date: 2000/12/11 16:38:12 $Revision: r11.11/1 PATCH_11.11 (PHKL
_23196)
@(#) dev_config.c $Date: 2001/05/18 16:19:34 $Revision: r11.11/1 PATCH_11.11 (PH
KL_24163)
@(#) core_cdio.c $Date: 2003/11/13 10:33:06 $Revision: r11.11/1 PATCH_11.11 (PHK
L_30044)
@(#) chassis_log.c $Date: 2001/05/01 10:27:46 $Revision: r11.11/2 PATCH_11.11 (P
HKL_24046)
@(#) aio_subr.c $Date: 2004/01/19 21:36:06 $Revision: r11.11/1 PATCH_11.11 (PHKL
_30317)
@(#) $Revision: libio.a: CUPI80_BL2000_1106_2 Mon Nov 6 15:44:21 PST 2000 $
sh /stand/build=>strings /usr/conf/lib/libfs-pdk.a | grep Revision
@(#) ufs_mchdep.c $Date: 2003/09/27 09:49:32 $Revision: r11.11/4 PATCH_11.11 (PH
KL_29527)
@(#) bcvalloc.c $Date: 2002/01/30 13:53:59 $Revision: r11.11/1 PATCH_11.11 (PHKL
_25886)
@(#) $Revision: libfs-pdk.a: CUPI80_BL2000_1028_1 Sat Oct 28 12:56:50 PDT 200
0 $
sh /stand/build=>

Ian

[Stephen Keane]

Worth a try, they are exactly the same versions as I have. Do you have the swagent log for the last PHKL_* patch that you applied?

[Ian Warner]

The last patch was a GOLDBASE11i I installed on tuesday to hopefully correct this error.

Ian (who is getting really desperate)

[Eric Antunes]

Ian,

Did you placed stape again at his place in /build/system?

Did you try to re-install the patch you removed??

Did you setted nflocks again to it's initial 2048 value???

Finaly, if none of this sugestions works, make a patch assement:

http://www5.itrc.hp.com/service/patch/assessSystemsPage.do

Eric