Hope I can explain this correctly..
Silverlight Application
Telerik Reports (in a separate Class)
Web appliation hosting the Silverlight — provides access to the reports via a ReportService.svc that points to the Telerik.Report.Server.dll.
Developed on my machine and all runs fine. There was one report that with certain amounts of data was timing out so via Google, etc… I managed to get the settings correct via httpBinding settings, IIS settings, and more and solved the problem. The report runs fine now.
When I deployed the application to the server and then hit the web application via my browser, all works just fine except again for the long report. The report works fine with less data, just times out and the service dies with larger data.
But when I run the same report on the server directly (using a browser on the server) everything works perfectly. Hit it from a different machine and everything is fine except the report with a lot of data.
So the ONLY difference in the two setups is that it works in a browser on the server, but not on my machine hitting that same server.
Any guesses on what could be the problem? It’s very very odd. Is it because the Silverlight application is actually running on the local machine and timing out?
The times we’re talking about are maybe 75 seconds to get a big report generated. Not several minutes or anything.
Any suggestions on what I could check or more information to provide? I could send all the settings, but since I figure they aren’t relevant since the only variable is what machine we’re using.
Odd…
UPDATE:
So… using Fiddler, I’ve looked at the calls from IE and Firefox. They are identical. The only difference is Firefox/Telerik closes the connection after about 30 seconds. The result still comes back, but Firefox is no longer listening. But if I look at the response from IE/Firefox they are the same. Either IE is keeping the connection open or Firefox is closing it prematurely. It’s not a firewall issue as it happens when everything is on the same machine.
I recently had to troubleshoot a problem which almost exactly matches your description: A WCF service which generates reports used to time out for larger reports. And changing the various WCF binding timeout values didn’t make any difference.
Only later I found out the the timeout actually happened on a network traffic manager (ZXTM in this case, but it doesn’t really matter). The host name that was used to access the service was actually registered with the traffic manager rather than the server directly, hence all traffic was routed through ZXTM. And that one happened to have a global 40 sec timeout defined.
A very effective quick-fix was then to use the server’s machine name in order to address the service.
Hence, I’d check how traffic to your server is being routed and if it might be a similar situation.