I'm running into an occasional native code error by JDK 1.4.1_01. The snippets that seem relevant are below, and the mentioned error log is attached(info). We're running on linux (2.4.*) and using Sun's VM.

Any ideas? This may well be related to the tomcat4 upgrade and associated mod_jk problems, and it is affected this wiki in a nasty way...

--ebu

Which libpthread version? which glibc version? for a similar bug with older jdks a workaround was ulimit -s 2048, limiting the stack to 2MB

Sorry, can't help you with this problem, but what is this tuparit-WebApp? The combination of libraries looks very interesting (though I am a bit surprised by the many log*-jars). -- Torsten

Torsten, it's a quick'n'ugly experimental setup for a housewarming invitation list using velocity and hypersqldb. Nothing you really want to see.;) --ebu

Signal 11's can be caused by a number of irritating factors - have a look at Sig 11 FAQ. Have you tried the installation on another box? If it's load related you might be able to cause it with a load tool like JMeter or WAST. -- Lachlan

It seems that we do not have this problem with JDK 1.3.1, only with JDK 1.4.1_01. Which does not really explain it any better :-). Installation is a bog-standard Debian Woody, with custom compiled Apache.

-- JanneJalkanen

OK, we still have this problem. Here's a list of things that we have tried.

JDK Tomcat mod_jk Result
1.3.1 3.2.3 pre-compiled JK1 OK.
1.3.1 4.1.18 pre-compiled JK2 OK (but clearly slower than 1.4)
1.4.1_01 4.1.18 pre-compiled JK2 fail
1.4.1_01 4.1.21 self-compiled JK2 fail
1.4.1 (Blackdown) 4.1.21 self-compiled JK2 fail
1.4.1_02 4.1.21 self-compiled JK1 Fail (but rarely)
1.4.2-beta 4.1.21 self-compiled JK1 OK?

We're currently trying the Sun Java 1.4.2-beta with 4.1.21 and self-compiled JK. I've set LD_ASSUME_KERNEL=2.2.5 for Tomcat - some of the things I googled up seem to suggest this.

I have a suspicion that this has something to do with the increasing load of the machine. Our libc is 2.2.5, pthreads 0.9.

-- JanneJalkanen

Update 10-May-2003: We've now been running 1.4.2-beta for a few days, and so far have had no problems. It could be that this was an infestation of bug 4785384 in the SDK. Let's cross our fingers and hope for the best.

Here are a couple of Java bugs that might have something to do with this:


An unexpected exception has been detected in native code outside the VM.
Unexpected Signal : 11 occurred at PC=0x40024C90
Function=(null)+0x40024C90
Library=/lib/libpthread.so.0

...

#
# The exception above was detected in native code outside the VM
#
# Java VM: Java HotSpot(TM) Client VM (1.4.1_01-b01 mixed mode)
#
# An error report file has been saved as hs_err_pid9331.log.
# Please refer to the file for further information.
#

Add new attachment

Only authorized users are allowed to upload new attachments.

List of attachments

Kind Attachment Name Size Version Date Modified Author Change note
log
hs_err_pid9331.log 20.0 kB 1 13-Feb-2003 11:45 Ebu
« This page (revision-20) was last changed on 10-May-2003 15:33 by JanneJalkanen