Re: FW: Hi , Semaphore problem --Pl. F1

You can do this in your kernel configuration. Look for the file called sem.h
in your kernel source code. (PATH_TO_LINUX_SOURCE/include/linux/sem.h). Then
you can change the values for SEMMNS (max semaphores), SEMMNI (max semaphore
sets),
SEMMSL (max number of semaphores per set). Then recompile your kernel. Please
remember to back up the original file first though :)

If your not comfortable hacking the kernel source, read up on it first. Oracle
recommends increasing the number of semaphores, so you could do some research
by looking some websites dedicated to using Oracle 8i on Linux for detailed
instructions.

(Yes, I see the irony of referring to Oracle docs, so please don't flame me
unless I'm really off-base here.)

If you need more detail, you may also do well with some maillists dedicated to
kernel modification. The only reason I know this stuff is because I had to
install Oracle once.

Moses wrote:

> HI,
> My name is Moses, I work as system admin in INDIA and I just got your ID
> when I was looking out for a solution on Semaphore Problem which I am facing
> currently. I will be greateful to u if you could help me out on this.
>
> I work on SCO Openserve 5.0 .. I have Progress loaded on it .. When ever
> I try to start more than 3 database in multiuser mode, it gives and error
> saying that "Semaphore limit exceeded" and terminates the process. therefore
> I am able to start only 3 databases at any given point of time. Can you help
> me out on this of how to increase the semaphore limit, I have already tried
> using ./configure command and increasing the values in the semaphore option
> but in vain... Can u help me on this ? ..
>
> Looking forward for your help.
> Thanks and regards,
> Moses.
>
> ================================================================
> --- Demo Copy ---
>
> Sent using "The PostMaster" by QuantumLink Communications
>
> One Internet account, unlimited personal e-mail addresses
> Get your free copy of "The PostMaster" at http://www.qlcomm.com/
> ================================================================