The problem occured on a dedicated production server, it never ever occured on a development server.
The server runs 4 websites: a "normal" customer website, two redirection websites whose only goal is to redirected the user to the normal website and a webservices website. All these websites are part of one Maven project, divided into 5 submodules. - a core module providing generic application services as well as data related services - a core-web module providing the normal customer website - a core-redirect module consisting of only one single JSP page for redirection - a catalogue-web module consisting of one single Tapestry page (just to show a difference frontpage) and some custom redirect logic - and, a ws module providing Spring Web Services based web services. These webservice are primarily used with a custom Flex application. The core module has a Maven dependency to Hibernate Entitymanager (version 3.3.1.javafabric, a custom made pom consisting of Hibernate 3.2.5.GA, Hibernate Annotations 3.3.0 and Hibernate Entitymanager 3.3.1). And there is also a Maven dependency to Spring (version 2.0.7). All webapplications, excluding the single page JSP redirection, have dependency on this core module. I am using Tapestry 4.1.3 for the two Tapestry submodules. Hibernate Entitymanager (or some of its submodules) has a dependency on Javassist version 3.3, Tapestry Project 4.1.3 has a dependency on Javassist 3.4.GA. All webapplications are deployed using (in Jetty terms) WebAppContext. (I am deploying 4 ROOT.war files in 4 difference directories). The server was under a heavy load when the problem occured. If my memory serves me well, I think the CPU had zero idle time and the general server load was about 4. All these numbers come from the 'top' util. This morning I checked the server log, generated by AWStats. During the evening hours the server had about 3GB data traffic per hour. Perhaps it is possible that the server was not able to handle such load. On the other hand, 3GB per hour is about 1MB per second. That isn't that much, is it? (My main expertise is Java. I am able to deal with basic Linux administration, but I am unable to determine if such a load is exceptional high.) Last but not least, some deployment settings. I am using Jetty version 6.1.5along with Java 1.5.0_14-b03 on a Linux Fedora Core 6 (linux kernel 2.6.22.7) server. The server has 2GB RAM and Jetty is started with '-Xms512M -Xmx768M'. Alongside with Jetty, PostgreSQL (8.1.9) is used as RDBMS. As said before, the problem never occured on a development server. My first thought when it happened was pure panic, my second thought this morning wanders about a serious server overload. I've googled already who to handle a Jetty server overload, but did not find anything. I will continue to google for a while and if nothing comes up, I will contact WebTide on this issue. Thank you for your time! Marcel
