Help me SYSTEM ERROR: (1083) ??

Posted by Admin on 30-Nov-2006 09:10

Hello everybody...

I'm running my PROGRESS 8.3a on Solaris 2.6 and go so good.

But i need to migrate my database on Solaris 8 (32bit)

when start my server it's running good, but..... when run any longer process like

a build application my server crash it and display me this error:

SYSTEM ERROR: Number of private slots in segment table exhausted(1083)

anyone can help me to solve this problem ???

I have this configuration on my /etc/system file and .pf file:

/etc/system:

*

  • Shared Memory Vars

*

set shmsys:shminfo_shmmax = 8000000

set shmsys:shminfo_shmmni = 256

set shmsys:shminfo_shmseg = 50

*

  • Semsys for Database

*

set semsys:seminfo_semmap = 512

set semsys:seminfo_semmni = 120

set semsys:seminfo_semmns = 1200

set semsys:seminfo_semmnu = 500

set semsys:seminfo_semmsl = 300

set semsys:seminfo_semume = 200

Start my server with:

_mprosrv -pf /produccion/lib/xponent.pf

/produccion/lib/xponent.pf:

  1. parameter for Xponent database

-db /databi_p/dbs/probanks

-ld xponent

-B 4000

-L 70000

-d dmy

-c 200

-T /tempo

-n 20

-yy 1950

This configuration it's ok on Solaris 2.6 but in Solaris 8 failed

help me please....

thanks..

P.D.

Sorry for my english, but is not my native language

All Replies

Posted by Admin on 30-Nov-2006 10:20

There are various similar entries in the knowledgebase with this platform. Basically, your new configuration is unsupported. You need to upgrade.

eg. KB P6546

Title: "Error 1083 with Progress 8.3B on Solaris" (Yes, I know you said 8.3a)

http://tinyurl.com/y6zb4w

There are also a couple of entries suggesting altering -B parameter, which you may like to experiment with first.

Your English seems ok, I understood you fine.

Message was edited by:

knuthandsome

Posted by gus on 30-Nov-2006 18:20

This error is a result of an unfortunate interaction between Solaris's memory and address space management policies (which changed in never Solaris releases) and the memory management policies used by the databasse. The old Progress address segment size (shared and private) is too small (16 megabytes). The segment size has been increased in later Progress releases.

One thing that /might/ help is to increse the value of shminfo_shmmax to 16,777,216 or larger.

But there aren't any known reliable workarounds.

It is not a good idea to be running software that is so old and then changing only parts of it because you will often run into incompatibilities amongs one thing or another.

8.3, although it was quite good, is a very old Progress release. So is Solaris 2.6 (released in 1997). The latest Progress release is 10.1A and 10.1B is due out in a few weeks.

You do realize that Solaris 8 is pretty old as well, don't you? (released in February 2000). The lastest version, Solaris 10, came out in January 2005, almost 2 years ago.

If you don't want to upgrade from 8.3 (although you should at least /consider/ 8.3E to get the bug fixes), then don't upgrade Solaris either.

-gus

Posted by Admin on 01-Dec-2006 08:44

Thanks,,,, knuthandsome

i will try to alterate -B parameter first...

have a nice day...!!!

()()()()o:

Posted by Admin on 01-Dec-2006 08:54

Thanks Gustav for your post....

i'm considering to upgrade my progress database..

Have a nice day.

()()()()o:

This thread is closed