Mapscript crashing Tomcat --> [Panic]

[Rodrigo Del C. Andrade]

    Dear Frank and Umberto:

    Thanks a lot for the advice, you really nailed it. I haven't tried 
reproducing it on mapserver cgi yet, but upgrading GD to 2.0.34 (I had 
2.0.33) apparently did the trick.
    I am testing it as I type, but it has been running for 20 min now 
and it haven't crashed yet, as opposed to the 20 secs crash.
    I tracked down that linux machine where it used to work fine and 
sure enough, that one had the newer GD in it.

    Thanks a whole lot.
    Sincerely

    Rodrigo

Umberto Nicoletti wrote:
> On 3/28/07, Frank Warmerdam <warmerdam at pobox.com> wrote:
>> Rodrigo Del C. Andrade wrote:
>> >      Greetings. dear list.
>> >      We are experiencing some troubling difficulties developing an
>> > application with mapscript on a Java web based environment with 
>> tomcat.
>> >    Here is the deal: while using the app (not nearly done, but
>> > functional enough at this stage) when one tries to zoom (in or out) or
>> > pan the map more than a few times, Tomcat often will go down screaming
>> > and the JVM just gives up on life.
>> >    I know this is probably not the right or best place to ask about 
>> it,
>> > but it seems to be caused by the interaction of mapscript and the GD
>> > library.
>> >         Here is the JVM output short version (long version is attached
>> > at the bottom):
>> ...
>> >    Any ideas? Full JVM error log is at the bottom. I am ready to 
>> give in
>> > more info if needed.
>> >    Thanks in advance;
>> ...
>>
>> > Stack: [0xb0203000,0xb0284000),  sp=0xb0279ed4,  free space=475k
>> > Native frames: (J=compiled Java code, j=interpreted, Vv=VM code,
>> > C=native code)
>> > C  [libgd.so.2+0x8182]
>> > C  [libgd.so.2+0x9a16]  gdImageLine+0x4a6
>> > C  [libmapscript.so+0xad355]
>> > C  [libmapscript.so+0xb310e]  msDrawLineSymbolGD+0xe7e
>> > C  [libmapscript.so+0xa75fb]  msDrawLineSymbol+0x8b
>> > C  [libmapscript.so+0xa9cb2]  msDrawShape+0x17e2
>> > C  [libmapscript.so+0xaa6d8]  msDrawVectorLayer+0x398
>> > C  [libmapscript.so+0xaa97a]  msDrawLayer+0xaa
>> > C  [libmapscript.so+0xaba67]  msDrawMap+0x6a7
>> > C  [libmapscript.so+0x552b5]
>>
>> Rodrigo,
>>
>> I'd suggesting seeing if you can reproduce this without Java 
>> involved.  I
>> suspect there is some sort of memory error in msDrawLineSymbolGD() 
>> (possibly
>> the infamous antialiasing bug?).
>
> Frank,
> I bet it is exactly that!
> Rodrigo: upgrade to gd 2.0.34 (http://www.libgd.org/Main_Page) and
> tell us if the problem goes away.
>
> Umberto
>
>>
>> Capture your map configuration into a map file, and run the map through
>> shp2img or mapserv with the appropriate layers turned on, and similar
>> view configuration *and* run it under "valgrind" to quickly highly
>> memory corruption as soon as it happens.
>>
>> If you can reproduce it this way,then you have something you can file
>> as a bug report.
>>
>> MapServer is not bug free, and it is important to be able to narrow
>> down bugs without too much extraneous stuff.
>>
>> Best regards,
>> -- 
>> ---------------------------------------+-------------------------------------- 
>>
>> I set the clouds in motion - turn up   | Frank Warmerdam, 
>> warmerdam at pobox.com
>> light and sound - activate the windows | http://pobox.com/~warmerdam
>> and watch the world go round - Rush    | President OSGeo, 
>> http://osgeo.org
>>
>
>