5631 2007-10-23 12:00 StackOverflowError processing WebMDS query 2008-06-25 16:41:33 1 1 1 Unclassified MDS4 general 4.0.4 All All NEW Teragrid P1 normal --- 1 navarro@mcs.anl.gov laura@isi.edu laura@isi.edu mdarcy@isi.edu neillm@mcs.anl.gov navarro@mcs.anl.gov 2007-10-23 12:00:08 This is actually MDS4 4.0.5. On RHEL4 on ia64. java version "1.4.2_14" Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_14-b05) Java HotSpot(TM) 64-Bit Server VM (build 1.4.2_14-b05, mixed mode) See fault in uploaded log @ 2007-10-01 10:39:06 navarro@mcs.anl.gov 2007-11-10 08:38:30 This problem is still happening, sometimes several times a week. A very interesting discovery this morning. When we get into the wedged state and WebMDS accesses MDS using a hosts external IP address (info=tgisinfo) everything works fine, BUT when it uses 127.0.0.1 (info=tgislocal) is gets this StackOverflowError. Restarting the MDS4 container makes this problem go away. Do WebMDS/Tomcat and/or MDS4 treat 127.0.0.1 differently? navarro@mcs.anl.gov 2007-12-18 12:28:57 I've now confirmed that does not have anything to do with using a normal host address vs 127.0.0.1. Whichever address is being used primarily between WebMDS and the TG WS MDS4 container can get in this wedged state. As of this instant. The following is an example of the error. http://info1.dyn.teragrid.org:8080/webmds/webmds?info=tgislocal&xmlSource.tgislocal.param.xpathQuery=//KitRegistration Attached is the failure output. laura@isi.edu 2008-01-07 01:42:35 Since attachments are disabled, I've added some relevant items (logs, debug output, etc.) to http://www.isi.edu/~laura/bug_5631/