Someone asked me to have a look at a Geoserver WMS which was showing strange 'white stripes'. Looking into this it appeared he wa
s using a Leaflet web app in epsg:3857, and serving a raster image which is actually a geotiff in epsg:28992 (later I found this behaviour in a totally unrelated ecw dataset).
When zoomed in enough, it seems as Geoserver is 'cutting off' part of the map image on the right side of the image. In QGIS zoomed in at about > 1:5000 see
it is seen on the right part of the mapcanvas.
But when you use a Leaflet application (which uses tiles) it is even more a problem because EVERY tile shows this cutoff aspect.
To make this more easy to view/reproduce, I created a little demo application:
which serves the same (downloadable) geotiff both via Geoserver and QGIS to see the difference.
Zooming in more almost removes half of a tile.
The tiff is downloadable and used style I will attach.
Note that when viewing this data in epsg:28992 (the native crs) all is fine! So it has something to do with (maybe?) a wrong clipping order or so?
I already answered this one above did you try jai-ext? We cannot fix jai.
Yep, geoserver there is now running jai-ext.
But the issue is still showing there (Geoserver 2.14.0)?
I cannot do anything for 2.14.0, it's already in the past... can you try at least with a nightly build of the 2.14.x series?
Duh..... 2.15 Snapshot ... working BOTH with and without jai-ext enabled!!!
Then downloaded 2.14.1 ... and also worked BOTH ways !! \o/
http://duif.net/cutoff/ is now running 2.14 with jai (NOT jai-ext)
So that was an easy fix
Thanks Andrea for your time, and sorry for not testing newest version earlier.
Assuming it has been fixed in 2.14.1 then