RV: [mapserver-users] MapServer consuming 100% CPU

[Fuchs Juan Carlos]

-----Mensaje original-----
De: Fuchs Juan Carlos 
Enviado el: Miércoles, 27 de Febrero de 2008 04:07 p.m.
Para: 'Desarrollo Desarrollo'
CC: mapserver-users at lists.osgeo.org
Asunto: RE: [mapserver-users] MapServer consuming 100% CPU


Sergio:
One's more:
in my .map  ( http://200.51.91.231 <http://200.51.91.231> .)
I separated all lines/poligons from labels.
Say:
layer Rivers
Tipe Line
Group Aqua...
several classes
the first for the big and important ones
the second for the adyascent and permanent rivers
the last for the little ones
each one restricted with minscale  / maxscale
 
layer Names
Group Aqua
type Annotation...
same classes, same maxscale  /  minscale restrictions
just different type sizes for each one
 
Remember that Truetype is a bit slower that Bitmap, but it's worth the
prize!
 
The group statement is to turn on / off in only one call.
 
On the other side it let's you track down the render time for each layer
so you will have a better control.
 
I never tried the new toy 
ANGLE Follow 
Its supoused to follow the line path along the river !!
 
If we are struggling to beat down the render time, keep it simple
but if you want it to be nice ...  :)
 
In reference to your Windows+ASP/Linux+Apache question, in my experience
it's quite the same
diferences i made comes from intensive (really intensive) map file tunning!.
http://mapserver.gis.umn.edu/docs/howto/mapfiletuning
<http://mapserver.gis.umn.edu/docs/howto/mapfiletuning>  (It's is a must
read!)
 
happy testing!
JC.

-----Mensaje original-----
De: Desarrollo Desarrollo [mailto:devmapserver at gmail.com]
Enviado el: Miércoles, 27 de Febrero de 2008 08:39 a.m.
Para: Fuchs Juan Carlos
CC: mapserver-users at lists.osgeo.org
Asunto: Re: [mapserver-users] MapServer consuming 100% CPU


Hello!
Thank you for your answer. 
I think you are right, please take a look in this thread to the answer I've
post to Ritesh Ambastha, I've reproduce the three tests using PostGRE /
PostGIS and although is the fastest data access the labeling is still the
hard work to MapServer. 
I'm going to play with MINFEATURESIZE, MINDISTANCE and all the options that
Mapserver allows to use, and I'll report back here the results, but just one
thing more to ask, Is this the normal behavior of Mapserver?, I'm trying to
confirm that if you do the same test with mapserver, no matter if you are
running in linux or windows, I wish to know that my Mapserver has no bugs or
incorrect / old libraries that makes labeling be slower.

Thank you very much, best regards
Sergio


2008/2/26, Fuchs Juan Carlos < jfuch at indec.mecon.gov.ar
<mailto:jfuch at indec.mecon.gov.ar> >: 

Hi Sergio:
PostGis maybe a good solution, in terms of stability (means system do not
crash, so often :) ).
But in concern with line labeling, things become bad.
 
Hydrologic data are a bunch of lines with irregular shape, and very often
segmented.
I mean that the same river could have trillons of segments in his path, and
in the same 
manner, trillons of labels !!
 
Try with some restrictions to the labeling work, 
say not allowing to labeling features under 5 say 7 pixels

MINFEATURESIZE 	Minimum size a feature must be to be labeled. Given in
pixels. For line data the overall length of the displayed line is used, for
polygons features the smallest dimension of the bounding box is used. "Auto"
keyword tells MapServer to only label features that are larger than their
corresponding label. Available for cached labels only. 


Value: [integer|auto]


say not allowing to repeat labeling features in spaces of 20 say 25 pixels
 

MINDISTANCE 	Minimum distance between duplicate labels. Given in pixels. 


Value: [integer]


 
Sometimes a lit bit of map file tuning is necesary
 
Hope its helps, 
JC.

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/mapserver-users/attachments/20080227/4c7340e6/attachment-0001.html