On Wednesday 14 February 2007 08:05, [EMAIL PROTECTED] wrote: > I have a virtual server under my control which runs a perl web > application and now a Java Tomcat app which uses mail a fair bit. I'm > loking for something which will tell me how much memory the perl app is > taking versus the Tomcat app over a period of say 1 hour or so. > > Googling for 'memory profiler web applications' and things brings up > things that you use to find memory leaks in apps which I dont want. > Naturally top just gives me instantaneous values which don't mean much > when a web app is only getting a few hits a minute or even less. > Thats why I want to get an average over a few hours or so. > > Also I don't have Gnome or any gui thing on this server so it has to be > command line or a perl or bash or other program that can be run from > command line. Output to file would be perfect. > > Does anyone have suggestions? What do people here use for getting stats > on programs like this?
If you are finding it hard ... it's because there is no such thing! An app running consists of [the app]-usually small [shared libraries]-usually big. So 10 apps is a bit more than 1 app. (10*app + shared libraries) At any one instant memory-usage-on-a-system vs system+app1 and system+app2 is a broad fuzzy kind of general hint. EG an app may use vast amounts of mem that is never touched and is safely ignored in swap. So it would look like a memory pig but really be a very attractive (efficiency wise) program. EG kde is much more memory hungry than icewm. How much? well lots! Does it show? Depends on what else the system is doing! The metric you are seeking could easily be totally misleading. If you insist on doing this study the output of "top" until you are bored. UNIX is not like windows. A metric on one does not compare to the other. Best advice: lower the physical ram until your system swaps. compare apps. James -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
