Request sent to forwarder are sent in a synchronous way.
It seems that the handling of incoming request is also done in a sequential way.
DNS server can receive a lot of request, an async behaviour is needed in order to avoid timeout or delays for clients.
That's a know TODO item.
IMHO it's a required feature if we want to have internal DNS as the default DNS.
Being able to flood the internal DNS just with my own computer is not a good sign.
Created attachment 7603 [details]
The attached patch fixes the bug for me.
Seems to work, I'm putting it a bit under stress and update the bug but it seems to do the job pretty well.
I tested and stressed the code a bit earlier today.
It seems to work really well.
Under the circumstances dns failed to respond most of the time before, namely while running an exchange schema update, it preformed much better now. The requests that are sent to the forwarder are answered immediately most of the time. The requests to the own zone are timing out every other time roughly, but this must be due to the load on the sam.ldb.
Maybe the next but is "the ldb server is too synchronuous" (or to slow...)
Cheers - Michael
This patchset passed autobuild last night. Marking bug as resolved.