Hello ALL.

I got a system with IDS 9.30.UC6 on SCO Unixware 7.12.
Sometimes I get error in online.log and server hungs.
My error
-------------------------
09:36:24 Fatal error in ADM VP at mt.c:12134
09:36:24 Unexpected virtual processor termination, pid = 1839, exit =
0xb
09:36:25 PANIC: Attempting to bring system down
09:36:25 semctl: errno = 22
09:36:25 semctl: errno = 22
09:36:26 semctl: errno = 22
09:36:26 semctl: errno = 22
09:39:07 IBM Informix Dynamic Server Started.
09:39:22 Segment locked: addr=0xa000000, size=2204229632
09:39:24 Segment locked: addr=0xa000000, size=2204229632
-------------------------

I think that server does not have necessary number of semaphores.
I post my /etc/conf/cf.d/stune file
-------------------------
NUMREGPT 256
NUMSCOPT 32
DRV_SEGKMEM_BYTES 0x147A000
RSTCHOWN 0
CONSOLE_SECURITY 0
SHMMAX 3500000000
SHMSEG 15
SEMMNS 8192
SCORLIM 0xC700000
HCORLIM 0x7FFFFFFF
HDATLIM 0x7FFFFFFF
SDATLIM 0x7FFFFFFF
SVMMLIM 0x7FFFFFFF
HVMMLIM 0x7FFFFFFF
SFSZLIM 0x7FFFFFFF
HFSZLIM 0x7FFFFFFF
HFNOLIM 2048
SFNOLIM 1024
NPROC 12500
ARG_MAX 1048576
SHMMNI 300
MAXULWP 400
MAXUP 160
COREFILE_SECURE 0
SEMMNI 8192
NHBUF 1024
NBUF 3000
-------------------------

I post fragment from my onconfig file
-------------------------
NETTYPE tlitcp,5,60,NET # Configure poll thread(s) for nettype
NETTYPE ipcshm,3,150,CPU # Configure poll thread(s) for
nettype
-------------------------

I have SEMMNS=8192 and SEMMNI=8192. Is it right?
If these number is not enough, which numbers I must set for kernel?

Maybe you show me other way for solving of my problem.

Thank!

On Jun 19, 8:15 am, s...@mail.ru wrote:
> Hello ALL.
>
> I got a system with IDS 9.30.UC6 on SCO Unixware 7.12.
> Sometimes I get error in online.log and server hungs.
> My error
> -------------------------
> 09:36:24 Fatal error in ADM VP at mt.c:12134
> 09:36:24 Unexpected virtual processor termination, pid = 1839, exit =
> 0xb
> 09:36:25 PANIC: Attempting to bring system down
> 09:36:25 semctl: errno = 22
> 09:36:25 semctl: errno = 22
> 09:36:26 semctl: errno = 22
> 09:36:26 semctl: errno = 22
> 09:39:07 IBM Informix Dynamic Server Started.
> 09:39:22 Segment locked: addr=0xa000000, size=2204229632
> 09:39:24 Segment locked: addr=0xa000000, size=2204229632
> -------------------------
>
> I think that server does not have necessary number of semaphores.
> I post my /etc/conf/cf.d/stune file
> -------------------------
> NUMREGPT 256
> NUMSCOPT 32
> DRV_SEGKMEM_BYTES 0x147A000
> RSTCHOWN 0
> CONSOLE_SECURITY 0
> SHMMAX 3500000000
> SHMSEG 15
> SEMMNS 8192
> SCORLIM 0xC700000
> HCORLIM 0x7FFFFFFF
> HDATLIM 0x7FFFFFFF
> SDATLIM 0x7FFFFFFF
> SVMMLIM 0x7FFFFFFF
> HVMMLIM 0x7FFFFFFF
> SFSZLIM 0x7FFFFFFF
> HFSZLIM 0x7FFFFFFF
> HFNOLIM 2048
> SFNOLIM 1024
> NPROC 12500
> ARG_MAX 1048576
> SHMMNI 300
> MAXULWP 400
> MAXUP 160
> COREFILE_SECURE 0
> SEMMNI 8192
> NHBUF 1024
> NBUF 3000
> -------------------------
>
> I post fragment from my onconfig file
> -------------------------
> NETTYPE tlitcp,5,60,NET # Configure poll thread(s) for nettype
> NETTYPE ipcshm,3,150,CPU # Configure poll thread(s) for
> nettype
> -------------------------
>
> I have SEMMNS=8192 and SEMMNI=8192. Is it right?
> If these number is not enough, which numbers I must set for kernel?
>
> Maybe you show me other way for solving of my problem.
>
> Thank!


I do not have a direct solution for you, but I remember that with
Unixware it was very important to have all the OS patches installed.
You should check carefully which set of OS patches you need, install
them and then go to further troubleshooting, if troubles remain.
Regarding kernel parameters, check the machine notes for your version
of Informix.

To me Unixware seemed more modern and less stable OS than OpenServer.
One needed substantially more time to invest to make it stable.
OpenServer was, after adding appropriate release supplement, almost
out-of-box production ready. That was one of the reasons I am still
using OpenServer for some Informix instances. (I have one Unixware
database server, but it runs legacy Oracle database). I remember I was
not the only Informix user prefering OpenServer to Unixware.


Darko Krstic

How I can to see all installed patches in my Unixware?

On Jun 19, 8:46 am, s...@mail.ru wrote:
> How I can to see all installed patches in my Unixware?


I don't have a Unixware box handy :-( It is rather pure SVR4, so you
can try pkginfo.

You say that you have UnixWare 7.12. I suppose it is in fact 7.1.2
(aka OpenUnix 8.0). I've never seen that one in practice. At location
http://www.sco.com/support/ftplists/ou8list.html you can get latest
(and final, since that product is being retired) maintenance pack for
your version of UW.

Darko Krstic

s_c_o@mail.ru schrieb:
> How I can to see all installed patches in my Unixware?
>
"pkginfo -c patch" gives a list of all installed patches.

Although we're not running 9.30 but 7.31 here's the
configuration for our UW 7.1.1 box which runs fine for
several years now:

NUMREGPT 256
NUMSCOPT 32
RSTCHOWN 0
SFSZLIM 0x7FFFFFFF
HFSZLIM 0x7FFFFFFF
CONSOLE_SECURITY 1
USER_RDTSC 1
SEGKMEM_BYTES 0x800000
SHMSEG 15
SEMMNS 2500
SEMMSL 300
SEMMNI 1000
SHMMAX 12000000000
SVMMLIM 0x7FFFFFFF
HVMMLIM 0x7FFFFFFF

SFNOLIM 2048
HDATLIM 0x10000000
SDATLIM 0x10000000
HSTKLIM 0x10000000
SSTKLIM 0x10000000
MAXUP 2000
NPROC 2000
SHMMNI 1000
FLCKREC 1000

onconfig:
NETTYPE ipcshm,7,64,CPU
NETTYPE tlitcp,1,128,NET
NETTYPE sqlmux

--
Roland Wintgen (Systemadministrator)

rw@evg.de
Tel. +49 2166 5508-23
Fax +49 2166 5508-20

EVG Elektro-Vertriebs-Gesellschaft Martens GmbH & Co. KG
Trompeterallee 244-246, 41189 Mönchengladbach
Tel. +49 2166 5508-0, Fax +49 2166 5508-90
www.evg.de info@evg.de

Sitz: Mönchengladbach, Rechtsform: KG, MG HRA 9281
Komplementär: Martens & Hamacher Verwaltungsgesellschaft mbH
Sitz: Mönchengladbach, MG HRB 2523
Geschäftsführer: Jörg R. Martens, Axel Hamacher, Dirk Martens-Ritz