Hi, all... This bothers me. I'm seeing this a lot on my 1.4-based server
at work. Right now it runs a bunch of fairly memory-and-processor-intensive
Perl scripts, light name service, and a couple other random, inconsequential
things.

Here's what top says a bit after the system becomes responsive again:

load averages:  0.19,  0.18,  0.24                                     12:24:55
38 processes:  1 running, 37 sleeping
CPU states:  0.0% user,  0.0% nice,  0.0% system,  0.0% interrupt,  100% idle
Memory: 11M Act 7116K Inact 328K Wired 213M Free 4K Swap 384M Swap free

The system has a lot of memory, and I don't see it dipping into swap since
it never seems to use enough to get there. FWIW.


Here's the /var/log/messages bit:

May 19 12:11:20 satserv1 /netbsd: WARNING: mclpool limit reached; increase NMBCLUSTERS
May 19 12:12:21 satserv1 /netbsd: WARNING: mclpool limit reached; increase NMBCLUSTERS
May 19 12:14:22 satserv1 last message repeated 2 times

I didn't have "options GATEWAY" defined, so I guess the NMBCLUSTERS was set to
the default 256. Yesterday on seeing this message again, I built a new kernel
with "options NMBCLUSTERS=1024" set. Evidently this wasn't enough...

One problem is that I can't do anything while the system is semi-wedged. It
seems to free up again in time, but by then I can't get useful information.
If someone wants to give me an idea about getting something worthwhile through
the debugger, I can try that...

Thanks in advance. I'm a tad nervous about this, as I've taken it upon myself
to inject NetBSD into the environment here, and this machine is going to be
at the center of the network before long.

Machine profile:

i386 (PII/233), 256MB RAM, NetBSD 1.4.

-- 
                           /\
     Mason Loring Bliss   /()\   mason@acheron.middleboro.ma.us
 awake ? sleep : dream;  /    \  http://acheron.ne.mediaone.net