[dotNetRDF-Support] "underlying connection was closed" errors when talking to Stardog

[Ron M. Z. <ro...@ze...>]

I've just recently started getting these weird "The underlying connection
was closed: A connection that was expected to be kept alive was closed by
the server" errors when talking to a Stardog database. Here are the
particulars:

   - Mostly happens when the Stardog database is on a remote server - in
   this case when I'm at home and querying a Windows 2012 server hosted at
   Azure. When talking to the database locally from the same box (at Azure) it
   rarely exhibits this behavior.
   - Same thing happens if I query Stardog on a Linux server at Azure
   - If I use a proxy between dotNetRDF and Stardog - like Fiddler - to
   examine what's going on, it always works! But stop using the proxy and it
   fails. Which makes it hard to diagnose.
   - It only seems to fail when the SPARQL query is long and a POST is
   necessary. GETs for query queries don't seem to fail.

After playing around and Googling for a while, I found this post (
http://stackoverflow.com/questions/13188058/httpwebrequest-the-underlying-connection-was-closed)
that suggested setting these options in the WebRequest:

request.KeepAlive = false;
request.ProtocolVersion = HttpVersion.Version10;

I added this to the CreateRequest function in BaseStardogConnector and it
seems to have fixed the problem!

So three questions:

1- Anyone have any ideas why this problem would seem to start for no
reason, after it'd been working fine for months?

2- Any idea why proxying through Fiddler would hide the problem? Shouldn't
it be passing the request straight through?

3- Is there any reason we shouldn't merge these changes into the codebase?


- Ron