performance degradation

[Reinhard Doelz]

( was :Re: Experiences with 4D/2xx as timesharing systems? )
>Authorizing-Users: Mark Callow <sgi!msc%ramoth.SGI.COM at ucbvax.berkeley.edu>
>
>If 4Sight was the problem (and it was a problem on early versions of 3.0
>due to memory leaks) it would be cleared up by logging out and
>logging in again.  If a reboot was necessary something else was (is?)
>wrong.

We have the same experience: Once the pidmax of 30000 cycles through one 
or two times (which ist the case on our 4D/80 every two weeks) the system
performance significantly degrades due to problems with the cron, SYSLOG
reporting the insufficient memory of ghost processes which do no longer exist,
4DDN blocking the console, pseude-terminals which do belong to a user but 
not to a session (/etc/whodo reports it), increase of swapping time (if you 
habe a large process in the back and some users attached, poor 4Sight does
no longer react promptly because it permits the big one to get swapped in)
etc etc...
The trouble is that the degradation is not directly measurable because you
won't force me to write down 30000 processes in order to make the bug 
reproducable.

Reinhard 

  
  ************************************************************************
  *   Dr. Reinhard Doelz           *           SWITZERLAND               *
  *     Biocomputing               *                                     *
  *      Biozentrum                * doelz%urz.unibas.ch at relay.cs.net    *
  * Klingelbergstrasse 70          *                                     *
  *     CH-4056 Basel              *                                     *
  ************************************************************************