Re: system is loosing temp disk space ... - BACK AGAIN

Yeah, this seems like it's a call to IBM. The DB group PTF would have
little impact, I believe, because the problem is in IFS, not database. But
you never know what effects come from what causes, sometimes.

I've seen a behavior in C, where memory allocated is not released until the
program. And multiple allocations use more memory, not what has already
been allocated. And on the 400, memory allocation is actually done on disk
- CPU memory is just a mapping of disk in the single-level sotre.

I was able, in a test, to see gigabytes of disk used up ins hort order,
even after freeing, I thought, the allocated memory. This is probably by
design, to some extent, but scary things CAN happen.

I think that sometimes allocations are kept past their 'official'
deallocation, in case they are needed again. But IFS should have a way of
"really" releasing allocations.

Just my ramblings.

At 02:44 PM 4/11/02 +0200, you wrote:
>Had IBM on the phone again today.
>They looked at the system and reported our problem back
>to the laboratory.
>I will report back if there any news ...
>
>As first help I do schedule a restart of TCP-servers during the night, now.
>
>...  Philipp
>
>Philipp Rusch schrieb:
>
> > We THOUGHT we solved it:
> >

-snip-

> >
> > We installed C2050450, installed the HIPERs and the DB-group PTFs
> > and had only a little pause, because some people were out ....
> >
> > Any help welcome.
> >
> > Regards from Germany, Philipp Rusch
> >
> > _______________________________________________