BayesFusion Support Forum

More jvm crashes even after removing the node properties.


#0 0xf6fe87a2 in _dl_sysinfo_int80 () from /lib/ld-linux.so.2
#1 0xf6eb2825 in raise () from /lib/tls/libc.so.6
#2 0xf6eb4289 in abort () from /lib/tls/libc.so.6
#3 0xf6ee6cda in __libc_message () from /lib/tls/libc.so.6
#4 0xf6eed56f in ...

We can reproduce the problem at will and it is due to user properties on fault nodes.

Basically if the user properties are very long (several KB) there is a problem. We have had to remove all user properties on the model to see the behaviour. If it does not crash, then we have a workaround (not ...

Hi,

The diagnetwork objects are not shared across threads.
Each thread has it's own object.

Meanwhile, the jvm crashed again. A core just got produced.


[myuser@myhost home]$ gdb java core.20677
GNU gdb Red Hat Linux (6.3.0.0-1.159.el4rh)
Copyright 2004 Free Software Foundation, Inc.
GDB is free ...

Hi,

Looks like core-dumps were happening - we did not see them since an oc4j deamon was restarting the server. The ulimit was set for coredumpsize of 0. This has now been changed to see what happens with debug libjsmile.so (that you sent).

Without the core-dump the only message we see is something ...

Hi,

The core.25595 (at the top of this post) was from
Jul 18 2009 07:42
no pid file produced at this time

and the hs_err_pid23197.log was from
Jul 20 2009 10:03
no core produced at this time

Sorry, for the confusion.

If you don't mind, can you please email me at mohsin.beg@oracle.com when the ...

Unfortunately no, since this is a pre-production environment and all test and production environments are linux only. Can I turn something on my side. I can put any debug statements in my code if you want.

Btw here is the excerpt from jvm pid.log



#
# A fatal error has been detected by the Java ...

The crash was not seen during bursts (many parallel diag requests) but during low load activity (i.e. one diag request).

Also jsmile is responding to diag requests at random as they come in from our monitoring systems (i.e. not user initiated), hence is unpredictable at best.

Sincerely,

-Mohsin

I am using jsmile in an application server environment. Every couple of hours the application server jvm core dumps. I am not sharing/reusing DiagNetwork objects across threads. I am using the jsmile for linux released on 04.16.2009

Opening the jvm core in gdb reveals the below. Can anyone advise ...

Hi Marek,

Unfortunately, I am still able to reproduce the problem at will. Please see attached screenshots (in zip).

And my genie (running on MS Vista) version is "2.0.3092.0 built on 6/19/2008"

Sincerely,

-Mohsin

Hi,

I think I may be hitting a SMILE/Genie issue.

In the attached zip (opening attached model and the case file)
when I set
"t24:found" as the pursued fault (with current posterior of 0.89),
"o168: shows a diagnostic value of 0.01.

Now when I set "o168" = notFound, the posterior
of "t24:found ...

I can consistently reproduce a jvm crash while calling AnnealedMap.
The details of the jvm log are attached.
Any suggestions of what may be causing this ?

Sincerely,

-Mohsin


#
# An unexpected error has been detected by Java Runtime Environment:
#
# EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc ...