Strange... when I call the `MHD_get_daemon_info()` function it unlock the clients. :-/ This[1] example doesn't lock the clients. Do I need to trigger the `MHD_get_daemon_info()` always?
[1] http://pastebin.com/dfykB1US On Fri, Dec 4, 2015 at 9:21 PM, silvioprog <[email protected]> wrote: > Oops, I forgot to inform my environment. ^^' It is Windows 7 64 bits and > MHD 0.9.47 compiled with MinGW-w64. I don't know if it happens on Linux too. > > On Fri, Dec 4, 2015 at 9:02 PM, silvioprog <[email protected]> wrote: > >> Hello, >> >> I'm trying to use Keep-Alive in the `MHD_USE_THREAD_PER_CONNECTION` mode >> with the `MHD_OPTION_CONNECTION_TIMEOUT` option, however, when timeout is >> ends, it locks the client that can't get more responses from my MHD server. >> >> Is very easy to reproduce that, you just: >> >> . compile and run this demo[1] (hellobrowser.c modified); >> . make some simultaneous requests in the `http://localhost:8080` link; >> . wait five or ten more seconds; >> . try a new request in the `http://localhost:8080` link. (the client >> will locks) >> >> (the timeout 3 was purposeful to allow to reproduce the problem) >> >> It seems that the connection is closed when the thread die, yes it's the >> right behaviour, but it seems that the client still connected in a 'ghost' >> connection, trying new requests in that. Is there any callback to force a >> client disconnection? I can't use the `Connection: close` header because I >> want to use Keep-Alive. >> >> Notice: using only the `MHD_USE_SELECT_INTERNALLY` it works fine even >> with a timeout, the problem happens only when I declare the >> `MHD_USE_THREAD_PER_CONNECTION` option. >> >> [1] http://pastebin.com/LpmvH87b >> > -- Silvio Clécio
