SMP on OSR 5.0.4

[bigdog175]

I'm trying to add a second processor to my Dell PE2400 everything is fine until I add the SMP software I have loaded patches rs504c - oss469d and oss623a I can run it in single user maintenance mode, but when I try to do normal boot I get past NLM services started and the OS locks up error message Panic: lock timeout .... cpu1. Any suggestions? Thanks

 

[vistech1]

I had a similar problem on a home built dual processor system. The way I was able to correct the problem was to get matching cpu's. I had previously had 2 unmatched cpus (both PII 266.

Once I put matched cpu's in, No problems.

Can't say for sure if this is your problem, but take a look.

 

[bigdog175]

Thanks for your input sturgis, but the proc's are matched Dual PIII 650mhz 256cache.

 

[stanhubble]

A couple of possibles (although not 5.0.4):



-------- Original Message --------
From: Roberto Zini <fred@strhold.it>
Subject: Re: Lock timeout on SMP machine
Newsgroups: comp.unix.sco.misc,comp.databases.informix

Bryan Tonnet wrote:
>
> We have the following setup;
>
> SCO OS 5.0.5 with RS505A
> oss497c
> SMP 1.1.1Eb
>
> Informix IDS (Workgroup) 7.30
>
> on a
>
> Dual Pentium II 450
> Intel 440GX+ board
>
> About 4 times a day (seems like 100 today) we get;
>
> Lock timeout: caller= .. lock= .. owner= ..
> Unable to freeze processor 2, proceeding
>
> then panic and crash. This has only been in evidence since we started
> using out RDBMS server in earnest (Informix 7.3). Has anyone seen
> (and cured) this SCO panic, possibly in conjunction with the database
> engine ?
>
> TIA
>
> Bryan Tonnet
> bryan@printman.com.au

Bryan,

I seem to remember that an incorrect installation order
could drive to problems like the one you're experiencing
(or better, I've read some articles posted here stating
that if the installation order is not fully honored, you
could have SMP problems even with RS505A and OSS497C
installed).

From memory, the installation pattern should be :

- install the base OS
- install the SMP package but DO NOT RELINK (if possible)
or reboot
- apply patch RS505A
- relink + reboot
- apply patch OSS497C


Other possible from compaq

CUSTOMER ADVISORY: EU010523_CW01

REVISION: 0

ProLiant ML350 Running SCO OpenServer 5.0.6 Panics With Lock Timeout Message During Heavy I/O


--------------------------------------------------------------------------------

NOTICE: The information in this document, including products and software versions, is current as of the Release Date below. This document is subject to change without notice.


RELEASE DATE: July 2001

DESCRIPTION

A Compaq ProLiant ML350 server using the Compaq Integrated 64-Bit/66-MHz Dual Channel Wide Ultra3 SCSI Adapter and running SCO OpenServer 5.0.6 and Compaq Extended Feature Supplement (EFS) Version 5.38A will display the following panic message on the console when experiencing heavy stress on the adapter :

&quot;WARNING: ad160: Adapter1 error(ha=1 bus=0 id=11 lun=0)
ad160: Host adapter failure
PANIC: lock timeout: caller F0144F6
lock=F02CBCE8
owner= CPU2 &quot; < /EM> < /EM> < /EM> < /EM>



This issue generally occurs when both channels of the adapter are fully populated with hard drives.


SCOPE

Any Compaq ProLiant ML350 server using the Compaq Integrated Dual Channel Wide Ultra3 SCSI Adapter and running SCO OpenServer 5.0.6 and Compaq EFS Version 5.38A.


RESOLUTION


To work around this issue, change the tunable parameter maxlockspins in the etc/conf/pack.d/crllry/space.c to a larger number (the default is eight, the maximum is twenty and the minimum is two).



OPERATING SYSTEMS AFFECTED: SCO OpenServer 5

SUBSYSTEMS AFFECTED: Non-Hardware Specific

THIRD PARTY PRODUCTS AFFECTED: Does Not Apply .

DOWNLOADABLE FILES: Does Not Apply