High oversampling on raster cells with reproject can put a significant amount of load on GeoServer
Description
Environment
is blocked by
Activity
Andrea Aime February 15, 2017 at 11:47 AM
Mass closing all resolved issues not modified in the last 4 weeks
Andrea Aime November 18, 2016 at 10:08 AM
In terms of backporting, this one is made of many commits. The geotools tickets normally have one commit with the ticket number in the commit message, the geoserver side has several commits all referring to this ticket number (so they have to be all tracked down and backported)
Andrea Aime November 17, 2016 at 1:22 PM
Euf... actually 2.10.1 is due in December, and 2.9.3 now... still, it's a truckload of changes and the new required TileCache implementation did not see much testing in the wild.
Andrea Aime November 17, 2016 at 1:21 PM(edited)
Linked to related issues (fixed on master, too many changes too close to a release for backporting now), including https://github.com/geosolutions-it/jai-ext/issues/135
Will backport once 2.10.1 is out, and then eventually to 2.9.x (after some time to allow checking the many changes do not come with unforeseen side effects)
Michael Steigemann August 18, 2016 at 11:30 AM
Hello!
This problem still occurs in release 2.8.5. The geoserver just get's killed when zooming in very far (the CPU stays at 100% ). When will there be a bugfix?
Thanks for your reply.
Best regards,
Michael
Scenario:
One source pixel is so big that covers 100 times the current screen (high oversampling)
Reprojection means we read 10 more pixels per direction
Scale up to bring that one pixel to the target resolution
The result is a giant image