Return-Path: Delivered-To: apmail-xml-cocoon-dev-archive@xml.apache.org Received: (qmail 83175 invoked by uid 500); 8 Jun 2001 16:56:35 -0000 Mailing-List: contact cocoon-dev-help@xml.apache.org; run by ezmlm Precedence: bulk list-help: list-unsubscribe: list-post: Reply-To: cocoon-dev@xml.apache.org Delivered-To: mailing list cocoon-dev@xml.apache.org Received: (qmail 83106 invoked from network); 8 Jun 2001 16:56:32 -0000 Date: Fri, 8 Jun 2001 09:56:16 -0700 (PDT) From: Benjamin Franz To: cocoon-users@xml.apache.org cc: cocoon-dev@xml.apache.org Subject: Re: Linux and Tomcat crashes In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Spam-Rating: h31.sny.collab.net 1.6.2 0/1000/N On Fri, 8 Jun 2001, Matthew Langham wrote: > Slightly off-topic but perhaps you can help: > > We are having problems running Cocoon on Linux (Suse 7.1) under heavy load. > After certain periods Tomcat just crashes. From what we have been able to > find out so far this may be due to the Java version and/or Tomcat version. > It does not happen under Win* > > Can I get some feedback on which Java / Tomcat version you are using under > Linux - especially in high-load scenarios. I would also be interested in any > similar experiences you have made and if possible - how you solved it. We saw two failure modes here for failures that _especially_ occured under heavy load. The first was simply needing too much memory for a transofrm and was addressed by increasing the allocated heap space from the default 64M to 256M. The second appeared to possibly be threading related - essentially if you push a servlet engine (we tried three different engines, including TomCat) hard enough on Linux using the Blackdown Java JDK (1.2.2 or 1.3-FCS) it goes runaway, gobbles all available memory and crashes (this was repeated with XT as servlet, Xalan-J-2 as servlet and Cocoon 1.74 w/Xalan-j-1.x.). The failure mode _never_ appears to happen when XSLT transformers are invoked via a command line program (although I've on rare occasion had 'hung' java processes that simply stopped doing anything and had to be manually killed to remove them from the process table). We actually had to abandon using servlets for XSLT rendering because of this. We did not find any successful work arounds otherwise. -- Benjamin Franz "Premature optimization is the root of all evil in programming." ---C.A.R. Hoare --------------------------------------------------------------------- To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org For additional commands, email: cocoon-dev-help@xml.apache.org