Very Computer

We have SCO Openserver 5.0.0b on a Corollary CBUS machine with 4 PENTIUM
166 Mhz processors, 64 MB RAM, a RAID with 6 - 4GB Disks and a 3com 3c905
Fast Ethernet card.

After the problem of "WARNING : ip: spinning on PCB Fxxxxxx" that has been
solved thanks to FCO.DIAZ and Jean-Pierre Radley, now we are experiencing a
strange but serious problem.

On this machine we have installed a copy of Conetic C-BASE database rel 3.7
that act as our main database and we are using standard telnet and Xterm
session to connect.

When one of our clients disconnect from the machine without "logout" or
"^D" a big big CORE image ( about 200 MB ) is generated under the directory
where the application is installed, then the machine begin swapping and
paging when this happens and all the users are logged out.

What I checked is that this big big core only happens when the C-BASE menu
command has been in execution on the client session that disconnect.

The menu process then remain PPID 1 and the CORE is generated.

Can anyone please help me ??

I know that there is a core parameters on System V that allow to specify
the soft and hard limit of a core dump file that a process can create (
SCORLIM & HCORLIM ) but I cannot find any spec on Openserver 5.0.

Thank You in advance.

--
Paolo Palmisano
====================

Quote:>I know that there is a core parameters on System V that allow to specify
>the soft and hard limit of a core dump file that a process can create (
>SCORLIM & HCORLIM ) but I cannot find any spec on Openserver 5.0.

Just an ugly workaround.
--
Berlin                    Juergen Russ                           Germany

  ulimit -Sc 2000   # limit core files to 2000 512-byte blocks, about 1MB

However, I'd like you to try something else.  Go into the kernel
parameters and check the value of SEC_STOPIO.  Tell me what it is.  If
it's currently 1, set it to 0, relink, reboot, and see if the problem
goes away.  If SEC_STOPIO is already 0 then don't change it.  Either
way, tell me what it is currently set to.  I'm working on a problem
that's similar to what you're seeing, and has to do with an interaction
with stopio().

Quote:>Bela<

In the beginning, or at least on Sun, 23 Feb 1997 19:34:03 GMT,  Juergen Russ

ls -s /dev/null core
chmod +w core

--


 Calcasieu Lumber Co.                    Austin TX
  ---- "Ya jus' mash that butt'n, righ' jere"  ---

--
Sam Liddicott                   |   Nothing I say is to be attributed as |
Campbell Scientific Ltd.        | a company statement or representation. |
Campbell Park, 80 Hathern Road, *----------------------------------------+
Shepshed, Leic. United Kingdom. LE12 9AL        Phone: +44 (0) 1509 601141

Please, where I can find gcc binaries for install them in a SCO Open
Server 5? ( preferred ftp site over Skunk )

----

Oscar Santos
Galicia-Spain
---

Quote:>Adrian, no puedo localizarte.

I typically never link to /dev/null but just make another node
in the proper directory with same major/minor as /dev/null.

However - putting a file called 'core' to intercept the dumps
will only work if the core is dumped in the same directory each
time.  Otherwise you have to put this workaround in place
anywhere that core is likely to dump.  This pre-supposes you
will never want to save that image.   Too often these
workarounds surprise you - such as the day something else dumps
core and you want to see the image - but alas it is gone.

--

: Please, where I can find gcc binaries for install them in a SCO Open
: Server 5? ( preferred ftp site over Skunk )

Depending on whether I am using the CD-ROM for my own purposes or not,
Skunk5 is available by anonymous ftp from kuso.shef.ac.uk.

--
Earl H. Kinmonth, Centre for Japanese Studies, University of Sheffield,