[MAPSERVER-USERS] msProcessProjection(): Projection library error.

[rich.fromm]

rich.fromm wrote:
> 
> 
> rich.fromm wrote:
>> 
>> 
>> rich.fromm wrote:
>>> 
>>> I am in the process of running another test, this using a build of the
>>> tip of branch-5-0, along with local patches both for bug 2497 and for
>>> bug 2533 (in other words, this patch).
>>> 
>> 
>> It also ran fine (for a 6 hour test under heavy load).  For more details,
>> see:
>> http://trac.osgeo.org/mapserver/ticket/2533#comment:6
>> 
> 
> I continued running stress tests on the branch-5-0 build.  While I have
> not yet seen the projection error again, after 06:24:13 elapsed time,
> tomcat again crashed in mapObj.new().  Note that this is very similar to
> the stack trace that I had seen previously (see
> http://www.nabble.com/msProcessProjection%28%29%3A-Projection-library-error.-to15815968.html),
> but not precisely the same.  The differences might be explained by this
> build being from branch-5-0, and the previous runs being from the 5.0.0
> release.  Regardless, I once again return to my notion that these are two
> separate issues:
> 
> _IO_fread (/lib/libc.so.6)
> yy_get_next_buffer (maplexer.c:3526)
> msyylex (maplexer.c:3357)
> loadClass (mapfile.c:2102)
> loadLayer (mapfile.c:2577)
> loadMapInternal (mapfile.c:4348)
> msLoadMap (mapfile.c:4573)
> new_mapObj (mapscript/java/mapscript_wrap.c:1612)
> Java_edu_umn_gis_mapscript_mapscriptJNI_new_1mapObj
> (mapscript/java/mapscript_wrap.c:19783)
> 
> 

Just wanted to give a little update here...

After on a few occasions (admittedly rare) seeing mapserver crashes in java
mapscript calls to mapObj.new() (discussed earlier in this thread) while
under
heavy load, I decided to to wrap the call to new mapObj(String filename) in
a
synchronized() {} block in my java code.  This did not have a significant
impact on performance, and I have not again seen the crash, even after long
(24 hours) stress tests.

While this doesn't prove anything, it is my belief that there is probably
some
threading/locking/synchronization issue in mapObj.new(), and that this is
separate from what has already been reported as bug 2533
(http://trac.osgeo.org/mapserver/ticket/2533), and that in the absence of
any
other bugfix, that this is a reasonable workaround.

I will file this as a separate bug.

- Rich
-- 
View this message in context: http://www.nabble.com/msProcessProjection%28%29%3A-Projection-library-error.-tp15815968p16112206.html
Sent from the Mapserver - User mailing list archive at Nabble.com.