mibiisa is part of the snmp thingys, my guess is that it is some remote
system monitoring tool but I havn't really bothered to look it up.
I see no reason to use it on a standalone machine anyway so I just
Look in /etc/init.d/init.snmpdx how the if statement is built, make
sure that it isn't true or just delete it from the rc?.d directorys.
It might be that it fills up with events and noone ever reads from it.
Which means that there is a bug, it really shouldn't fill up all
memory and swap under any conditions.
Life - the ultimate practical joke
I've noticed the following behavior on a handful of our Solaris 2.6
machines (I don't think these are running a recent cluster patch). When
snmpdx is started, mibiisa starts consuming more and more memory, sometimes
100 MB in-core, or more. I don't recall seeing any patches for this on
Sunsolve, but I was wondering a) if anyone else had noticed this problem and
b) what you did to clear it, if anything.
I haven't noticed any peculiar errors related to snmpdx/mibiisa popping up
in my logs, so this seems kind of mysterious.
The problem is particularly pronounced on two webservers, pushing a lot of
traffic and each has many virtual interfaces, however I have other Solaris
2.6 machines which move substantially more data across the network and their
SNMP agents are well-behaved wrt memory/swap consumption...