rcee <4mystudies@gmail.com> wrote:
> Please help me on this. I have some debates going on with the database
> team and OS admins , basically finding fault with each other. the
> database ( jBASE ) people says that it is a AIX problem where as the
> OS Admins says the against.
>
> After debugging the acitvity on the application i found that the delays
> being experienced in
> the semop() calls within in the jBASE Application .
>
> Is there any other way to confirm this , Any body knows more about the
> semaphore handliing on AIX 5.2 !
>
> When i do a DLPAR to the CPU;s even after adding the resources the
> throughput of the job remains the same. Why the job is not making use
> of the extra added resource.
>
> Is there any parameters to be set for this !

semop() is a function that is supposed to create delays, to some extent
that is its function: make a process sleep until a certain condition
is satisfied, a resource is available.

I understand that your problem is that a program using a jBASE database
is not running fast enough, right?

I would say that the delays in semop() are not your problem - maybe the
problem is what resource is so busy that processes have to wait for it,
thereby delaying the whole program?

Database and application tuning is a difficult thing that can certainly
not be done remotely in a newsgroup. You will have to get the database
people and the AIX people to work together to figure out where the
bottleneck is.

Excessive OS resource consumption can be caused by a badly configured
database, and bad database performance can be caused by a misconfigured
operating system. You have to look at both together.

Yours,
Laurenz Albe