At 19:03 +0200 12-04-2005, Michiel Meeuwissen wrote:
Henk Hangyi wrote:
 Hi,

 I have an MMBase website which runs on the following configuration.

 JDK 1.4.1_07-b02
 Tomcat 4.1.31 (making use of AJP13)
 MMBase 1.7.3
 Struts
> OSCache 2.1

Is OSCache set to a maximum? Does it cache to memory and/or disk?

>
 The website works perfect ... untill the garbage collector starts. After
 the gc starts the website hangs for a couple of minutes. When the gc runs
 the load on the server remains low and there are no Java processes taking
 a lot of CPU.

 Did anybody see this behavior before? Any guess about what could possibly
 be the cause and how to fix it?

I have no idea, but I think ussually extremely long gc's indicate that the memory is nearly exhausted. Did you switch on gc-logging? I suppose

And of course: long gc's occur on large memory settings. Is your JVM set to 1 Gb or more? My experience is that a lot of MMBase sites run perfectly at 512 Mb or less, even complicated, big clouds. And i to, would advise you to log the gc's to get some picture: -Xloggc:/path/gc.log.
Big memory settings keep relatively long to fill up (with MMBase node caches, oscaches etc.). One of our sites was set on 1,5 Gb and could take sometimes 2 minutes to garbage collect, before we tuned it. For more information: http://java.sun.com/docs/hotspot/VMOptions.html


And read Nico's post and its thread in this list of May 8th last year.

---Andr�
--
Andr� van Toly
http://www.toly.nl                                mobile +31(0)627233562
------------------------------------------------------------------>><<--
_______________________________________________
Developers mailing list
[email protected]
http://lists.mmbase.org/mailman/listinfo/developers

Reply via email to