Originally posted by: SystemAdmin

All,

I have a vendor that is kicking out a ton of defunct (ZOMBIE) processes. From some documentation it reads, "Such and application, if long running, can with time fill up the process table to a unacceptable level".

Does anyone know how to quantify an unacceptable level?

Thanks!
#AIX-Forum

Originally posted by: SystemAdmin

<HINT>
In a former position,
quantifying process table saturation,
and determining thresholds came up in connection with
remote reporting of server health.

Among the customers,
there was violent disagreement as to both metric and threshold.
There was agreement as to symptom to be avoided, "thrashing",
an intermittent cessation of external response,
and narrow window of allowed external stimulus.

The three major metrics which had to be implemented
quantified the process table saturation as a percentage of
<cumulative userspace time>/(<clock time> - <watchdog/scheduler/idle time>)
or
<cumulative userspace+system time>/<clock time>
or
<cumulative userspace+system time>/(<clock time>-<watchdog/scheduler/idletime>)

IIRC, The acceptable threshold for the first was about 90%,
about 70% for the second, and about 100% for the third.
</HINT>
#AIX-Forum

Originally posted by: nagger

See answer there, please.
#AIX-Forum