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

[rich.fromm]

rich.fromm wrote:
> 
> Bug 2533:
> http://trac.osgeo.org/mapserver/ticket/2533
> 
> As I noted in the bug, I can test any fix in my local environment.
> 

I am in the process of running a test with the code from the trunk including
this change, plus a patch for bug 2497.  It can take several hours for
problems to be reported, so my debugging cycle is a little slow here.

Even if this solves the projection errors reported in msProcessProjection(),
I
suspect there still might be a separate error causing a crash.  Earlier I
reported a crash and was hypothesizing that they were related, but I now
have
more evidence to suggest that this may not be the case.

Last night I ran a test that did NOT include the patch for bug 2533. 
Rather,
the codebase was the 5.0.0 release, plus patches for bug 2525 and bug 2497.

On several occasions there were variants of the error reported earlier here.
I believe my retry logic worked (although unfortunately I again screwed up
the
logging, so I don't have as much detail as I would have liked), but tomcat
did
not crash, and the test continued:

Mon Mar 3 19:59:46.208 PST 2008
java.lang.UnknownError: msProcessProjection(): Projection library error. 
	at edu.umn.gis.mapscript.mapscriptJNI.new_mapObj(Native Method)
	at edu.umn.gis.mapscript.mapObj.<init>(mapObj.java:288)

Mon Mar 3 20:28:28.951 PST 2008
java.lang.UnknownError: msProcessProjection(): Projection library error. no
system list, errno: 22

	at edu.umn.gis.mapscript.mapscriptJNI.new_mapObj(Native Method)
	at edu.umn.gis.mapscript.mapObj.<init>(mapObj.java:288)

Mon Mar 3 20:57:45.000 PST 2008
java.lang.UnknownError: msProcessProjection(): Projection library error. 
	at edu.umn.gis.mapscript.mapscriptJNI.new_mapObj(Native Method)
	at edu.umn.gis.mapscript.mapObj.<init>(mapObj.java:288)

However, at 2008-03-03 21:08:55 -0800, tomcat crashed, with the following
stack trace (previously seen and reported earlier in this mailing list
thread):

_IO_fread (/lib/libc.so.6)
yy_get_next_buffer (maplexer.c:3522)
msyylex (maplexer.c:3353)
loadLayer (mapfile.c:2572)
loadMapInternal (mapfile.c:4353)
msLoadMap (mapfile.c:4578)
new_mapObj (mapscript/java/mapscript_wrap.c:1612)
Java_edu_umn_gis_mapscript_mapscriptJNI_new_1mapObj
(mapscript/java/mapscript_wrap.c:19633)

There's no immediate sign of anything going wrong right before the crash.

The only other sign of something going wrong is the following earlier msg
from
the tomcat log:

Mar 3, 2008 8:38:19 PM org.apache.tomcat.util.threads.ThreadPool logFull
SEVERE: All threads (600) are currently busy, waiting. Increase maxThreads
(600) or check the servlet status

But that was a while before, so I'm not sure how much worry to put into it.
Although I understand that tomcat has been known to sometimes get into a bad
state after outputting that message, but it did run for quite some time
afterwards with no visible problems.

I'm not sure whether the fact that both the reported projection error and
the
crash were from mapObj.new() is significant.  Perhaps the fix for the
projection error (if it works) will also address the crash, perhaps not. 
I'll
report as soon as I have any further data.

If all else fails, I could try a synchronized() block around the
mapObj.new()
call in the java code as a workaround.  Although I'd rather not resort to
such
a coarse-grained solution, but it being only a single line of java code (and
presumably not nearly as time consuming as something like mapObj.draw()),
maybe it wouldn't have that serious of an impact on performance.

- Rich

-- 
View this message in context: http://www.nabble.com/msProcessProjection%28%29%3A-Projection-library-error.-tp15815968p15837796.html
Sent from the Mapserver - User mailing list archive at Nabble.com.