From: trickard <nu...@jb...> - 2005-03-07 19:11:49
|
We are running JBoss3.2.5 on HPUX11.x. We upgraded last fall from 2.4.3 and have been running smoothly since the upgrade. We ran without problems for several years before the upgrade. We recently made a small change to a web application and two ejbs that the web app uses. The webapp calls a new method added to one of the beans and this new method calls a new method added to the second bean. The ear file has 100 or so beans. The ear was updated with the new files using the sun deploytool and redeployed to the dev and then tst environments. Under some circumstances manipulating the web app causes the jboss jvm process to totally disappear. There are no messages written to the console or any log and there is no core dump for the process the PID just vanishes without a trace. The failure typically happens on an update in the web app and the display of the returned page just stops at some point and the java process that was running the jboss jvm is simply gone. The new code works without errors until the jvm disappears. What could client code do that would cause the failure of tomcat/jboss in this way. I didnt think it was possible for client code to do anything, however wrong, that would immediately kill the jboss/tomcat jvm. We have had resource issues in the past but have always had failed threads announce their failure before dying and have never had anything kill tomcat or jboss. Is anyone aware of this behavior anywhere. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3869099#3869099 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3869099 |