Dear Sun-Managers,
a while ago I posted an inquiry on how to find swap-space-consuming
processes inside a package of routines.
I got several helpful answers. Here's a short summary:
a) Run and monitor SZ of ps -el
--> simple but not so comfortable
b) Use a (PD) program called "top" for monitoring. It's great.
It shows a list of processes, ranks them for cpu use and lists
the memory associated with each.
It really helped me tracking down the memory leak.
c) I lots of suggestions to use "purify".
As far as I know this is a commercial tool.
I haven't tried it.
d) There was on suggestion to use "proctool".
I haven't tried it, either.
Thanks to
Birger.Wathne@vest.sdata.no
mikebe@pass.bt.co.uk
dworrell@isnet.is.wfu.edu
hendefd@techserv.tech.duc.auburn.edu
unxsscivan066!unxsscivan066.sierrasys.com!rstar@wimsey.com
rruiz@spin.com.mx
tom@uni-paderborn.de
stern@sunrise.East.Sun.COM
jaanus@ml.ee
don@mars.dgrc.doc.ca
Geert.Touquet@uz.kuleuven.ac.be
Andreas
My original posting was:
========================
> I'm in severe trouble:
> ----------------------
>
> On a SS10, running Solaris 2.3, all computing activity of a program
> package ends in a deadlock after a certain amount of time.
> The reason is, as far as I have found out using 'sar', a permanent
> decreasing area of available swap space.
> After that the paging activity reaches it's peak and the machine locks up.
>
> Since I'm requested to find out which of the processes is the reason
> for the swap space consumption, I'd like to ask you if there is someone
> who can give me a piece of advice.
>
> Is there a way to tell which processes are to be associated with
> used swap space at a certain time (a kind of mapping) ?
>
> What is the best way to track down the reason for the deadlock ?
>
> Any comments, hints, experiences etc. are highly welcome.
>
>
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:09:17 CDT