I'm trying to figure out where I have memory pressue coming from in my WebSphere environment.
I have a WebSphere footprint of 839 megs, real memory of 1.2 gigs, VDISK of 120 meg and real swap of 830 megs (I know, a little short). The WebSphere environment seems to be fine during the day, when I should have users working on it. The problem appears to be happening later in the day. Over the weekend I had one linux guest generate 500 megs of swap space usage and VM paging itself was up tp 87% utilized. A couple other guests had 200-300 megs of swap generated. The one thing these three have in common seems to be xpolog going against an httpd exposed file directory. I would prefer to use samba, but the windows folks are reluctant to let Linux play in that sandbox because they mistakenly assume that my penguins are gonna try to be primary WINS servers. (further education required there) On the surface, it appears that the one of the two server regions as well as the deployment manager have a bunch of the memory that is swapped out allocated, simply by the change in the memory when components of WebSphere are shut down and restated. This is inconclusive because all that really indicates is that during the shutdown process, memory that was swapped out was swapped in and there was no pressure to swap out something else to handle it. In all cases, Cache was between 125 and 185 megs and buffers were between 77 and 107 megs. Free memory was running about 65 megs. I am wondering if anyone here has any sizing experience with WebSphere under Linux on z/Series, or insight into some memory monitor tool that I can use to say process X has x many pages real and x many pages swapped out? I would at the very least like to run something like a VM stat that runs for an hour at 10 min increments then logs a time stamp and goes again for x many hours. I have a script that will do that, but I was wondering if anyone had an idea on how to incorporate some aspect of top or another tool so I can try to figure out where the memory pressure is coming from. On the Non websphere guests this swap behaviour is not beeing seen. It's swapping maybe a meg in those systems. Usually less if at all. Thanks -J ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
