The original message:
> We often get the following message on a DEC 3000/300X :
> No more processes.
> and no command can be run during several minutes.
> There are on the average 250 running processes. The memory size is 64Mb,
> with 600Mb swap space (40% free).
> We upgraded from OSF 3.0 to 3.2 with no change.
> A sysconfig with max-proc-per-user = 256 did not change anything either.
> Note that there are also a lot (20) of <defunct> processes.
Thanks a lot to:
Pierre Wendling <wendling_at_vbo.dec.com>
Hellebo Knut <Knut.Hellebo_at_nho.hydro.com>
Alan Rollow <alan_at_nabeth.cxo.dec.com>
Gyula Szokoly <szgyula_at_skysrv.Pha.Jhu.EDU>
Hans de Hartog <ccult1!bommel!dehartog_at_relay.nl.net>
Kristian Koehntopp <kris_at_oekoalpha.pz-oekosys.uni-kiel.d400.de>
who indicated that the solution is to increase the "maxusers" parameter in
the kernel config file (/usr/sys/conf/machine_name), then config.
Kristian Koehntopp said that he found in param.c in /usr/sys/conf:
#if !defined(MAXPROC)
#define NPROC (20 + 8 * MAXUSERS)
#else
#define NPROC MAXPROC
#endif
Pierre Wendling found that the <defunct> processes were created by exmh
mailer because of a bug in a previous version of tcl/tk. I have upgraded to
tcl7.4p1/tk4.0p1 and defunct processes have vanished.
Thanks for the fast and precise answers.
| From: Frederic Arenou |--------------------------------------------------|
| At: DASGAL- CNRS URA 335 | e-mail: Frederic.Arenou_at_obspm.fr |
| Observatoire de Paris | URL:
http://www.obspm.fr/cgi-bin/whereis?=Arenou |
| F-92195 Meudon Cedex | Tel: (33 1) 45 07 78 49 / Fax: 45 07 78 78 |
Received on Fri Sep 22 1995 - 15:57:41 NZST