Strange mapscriptJNI_delete_1layerObj+0xf error killing Tomcat

[Rodrigo Del C. Andrade]

    I have some interesting news to report.
    First of all, thanks Umberto for fixing the make test error so 
quickly. You rock!
   
    Ok, onto bussiness. I found out that make threadtests was a 
wonderful way to reproduce the original error that was making tomcat 
crash, and probably why it only happened on the server and not on my 
machine. Running it with 1 thread (as it was being run in my single 
processor / single user machine), it wouldn't crash. But running the 
threadtests on the 4 processor server with 10 threads, it crashed 
consistently about 65% of the time, on a 300 runs test. I increased it 
to 30 threads and I got a 100% crash rate.
    Now, witth Svn rev 6167, the same 300 consecutive runs with 10 
threads of threadtests had a whooping ZERO crash rate!
   
    Just to be sure, I increased it to 100 threads, and am about to let 
it burn 10.000 times.  I cant see how the application will will fare on 
a real life situation yet, as apparently the folks responsible for 
generating new versions on the production server don't come to work on 
sundays. Slackers.
    I will report back in the morning.

    Best regards,

    Rodrigo
   
-- 
*Rodrigo Del C. Andrade*
/Programador/
/SIC - SSE - Soluções Segurança Pública/

*DÍGITRO TECNOLOGIA*
*E-mail:* rodrigo.andrade at digitro.com.br 
<mailto:rodrigo.andrade at digitro.com.br>
*Fone:* +55 48 3281-7314 / +55 48 3281-7000
*Fax:* +55 48 3281-7299
*Site:* www.digitro.com <http://www.digitro.com>



Umberto Nicoletti wrote:
> On 6/1/07, Oliver Wesp <wesp at gdv.com> wrote:
>> Hi everybody,
>>
>> I just want thank everybody for the information in this thread. We're
>> facing similar issues with java mapscript crashing the vm. Unfortunately
>>   the application relies on the dynamic creation of layers and classes.
>>
>> I tried the svn version this afternoon and it seems to be stable. No
>> more tomcat crashes till now. I will do some more testing on monday but
>> it looks like Umberto was right with expecting some good news :-)
>
> Yes, party time! ;-)
>
> Umberto
>
>>
>> Just in case this information is usefull: We're running our application
>> on Solaris 8 using Shapefiles, GDAL Raster (HFA) and WMS as datasources.
>> Tomcat is  5.0.28, JAVA is j2sdk1.4.2_11.
>>
>> Regards,
>> Oliver
>>
>>
>>
>> Umberto Nicoletti schrieb:
>> > I suggest against this practice (no pun intended Benedikt, I sincerely
>> > appreciate your help).
>> > I say let's wait until Rodrigo tries out a rfc-24 enabled build,
>> > frankly I am quite positive rfc-24 will fix his issues.
>> >
>> > Regards,
>> > Umberto
>> >
>> > On 6/1/07, Benedikt Rothe <umn-ms at hydrotec.de> wrote:
>> >>
>> >> Hi
>> >>
>> >> Some developers (Umberto, Tamas) work on a solution of this problem.
>> >>
>> >> As a workaround I made good experiences with
>> >> http://lists.umn.edu/cgi-bin/wa?A2=ind0506&L=mapserver-users&P=87500
>> >>
>> >> > if we dynamically create mapserver-objects, we allway use code like
>> >> this:
>> >>  > classObj cO = new classObj(...);
>> >>  > ..
>> >>  > // Do something with cO
>> >>  > ..
>> >>  > //now cO isn't used any more
>> >>  > cO.delete()
>> >>
>> >> With "freeing manually" we made the Mapserver inside Tomcat quite 
>> stable.
>> >> Hope, this helps
>> >> Benedikt
>> >>
>> >>
>> >> UMN MapServer Users List <MAPSERVER-USERS at LISTS.UMN.EDU> schrieb am
>> >> 31.05.2007 15:13:41:
>> >>
>> >>  >
>> >>  >     Hello dear list.
>> >>  >
>> >>  >    A problem is happening in our production machine which we were
>> >>  > unable to reproduce on our development machines. We have a very
>> >>  > large application almost ready for deployment, in which one of the
>> >>  > modules is written in java mapscript 4.10.1,  and the error on the
>> >>  > attached log happened for the first time when we installed the
>> >>  > application on the production server. Whats worse, the GIS module
>> >>  > causing the error brings down Tomcat and the whole application 
>> with
>> >> it.
>> >>  >     This is a excerpt from the JVM log:
>> >>  >
>> >>  > # An unexpected error has been detected by HotSpot Virtual 
>> Machine:
>> >>  > #
>> >>  > #  SIGSEGV (0xb) at pc=0x636ff93b, pid=19162, tid=1759710128
>> >>  > #
>> >>  > # Java VM: Java HotSpot(TM) Server VM (1.5.0_07-b03 mixed mode)
>> >>  > # Problematic frame:
>> >>  > # C  [libmapscript.so+0x3893b]
>> >>  >
>> >> Java_edu_umn_gis_mapscript_mapscriptJNI_delete_1layerObj+0xf
>> >>  > #
>> >>  >
>