On 15 March 2017 at 20:17, Evgeny Grin <k...@yandex.ru> wrote:
> If you resume connection - it timeout timer will be restarted. So, right
> after resume connection should not timeout.
> [...]
> Sorry, still unclear for me, why do you need custom timeout connected
> with suspend/resume. And what is not working.
My current application logic is as described below:
0. We are inside DH, current connection is #0 (active uploader) and
there is second connection #1
1. DH #0 calls MHD_suspend_connection( #1 )
2. After #0 finished a work, DH #0 or Completed callback #0 calls
MHD_resume_connection( #1 )
3. You are thinking that MHD_resume_connection() will reset a timer.
But this is wrong, see below.
src/microhttpd/connection.c
2732 int
2733 MHD_connection_handle_idle (struct MHD_Connection *connection)
2734 {
...
3218 timeout = connection->connection_timeout;
3219 if ( (0 != timeout) &&
3220 (timeout <= (MHD_monotonic_sec_counter() -
connection->last_activity)) )
3221 {
3222 MHD_connection_close_ (connection,
3223 MHD_REQUEST_TERMINATED_TIMEOUT_REACHED);
3224 connection->in_idle = false;
3225 return MHD_YES;
3226 }
3227 MHD_connection_update_event_loop_info (connection);
3228 #ifdef EPOLL_SUPPORT
3229 if (0 != (daemon->options & MHD_USE_EPOLL))
3230 return MHD_connection_epoll_update_ (connection);
3231 #endif /* EPOLL_SUPPORT */
3232 return MHD_YES;
3233 }
In my case the problem is on line 3220, more exactly this is a cause
of the problem:
timeout <= (MHD_monotonic_sec_counter() - connection->last_activity
A difference between git and 0.9.51, 0.9.52:
1. git version calls MHD_RequestCompletedCallback with
MHD_REQUEST_TERMINATED_TIMEOUT_REACHED (and it is cool).
2. non-git versions don't do this, instead they send INTERNAL_ERROR
silently & immediately (don't know why).
I hope now it is clear.
> Why not to limit number of connection to 1 by
> MHD_OPTION_CONNECTION_LIMIT? Then you will process single connection
> only at any time.
> To force close connection after response, add "Connection: close" header.
Because there are also clients with GET requests, the server must send
responses to them.
Let's say, the server is working in some kind of a stupid database
mode: it provides read access to anyone and there is must be only one
writer.
> Patches are always welcome. If you improve MHD functionality and don't
> break anything - we will merge it.
I sent the patch already:
http://lists.gnu.org/archive/html/libmicrohttpd/2017-03/msg00031.html
There is typo in both notes (.h & .c): s/setting/getting/. Sorry, I missed it.
--
With Best Regards,
Vitaliy V. Tokarev
