Re: [Libevent-users] Re: libevent http and lost requests

[about bus <aboutbus@xxxxxxxxx>]

On Tue, Aug 16, 2011 at 4:51 AM, Mark Ellzey <mthomas@xxxxxxxxxx> wrote:

On Sun, Aug 14, 2011 at 04:15:10PM +0300, about bus wrote:
> I've checked this, domain name resolution does not take much time.
> Issue somewhere between end of evhttp_socket_cb and start of evhttp_read_cb
> functions calls.
>
> 2011.08.14 16:52:27 [warn] ===  evhttp_socket_cb: fd: 131
> 2011.08.14 16:52:27 [warn] ===  evhttp_get_request_connection: fd: 131 new
> request from 127.0.0.1:53282
> 2011.08.14 16:52:27 [warn] ===  evhttp_get_request_connection [[before
> return]]: fd: 131 ### return from function
> 2011.08.14 16:52:27 [warn] ===  evhttp_start_read: fd: 131
> 2011.08.14 16:52:27 [warn] ===  evhttp_socket_cb [[before return]]: fd: 131
> ### return from function
> 2011.08.14 16:52:57 [warn] ---  event_active: 0x803032390 (fd: 131), res 2,
> callback 0x800797140
> 2011.08.14 16:52:57 [warn] ===  evhttp_read_cb: fd: 131 len: 401: data: GET
> /qyery/xxx/yyy/ <-- headers --->

These logs are looking different each time you post. Are you able to
reproduce this in a consistent manner?

Yes, these logs are different, because I am making chages to check different situations and assumptions.
The logs above shows what there is no problem in DNS resolving, execution time of evhttp_get_request_connection less than 1 second.
Yes, I can reproduce this in a consistent manner, I am always checking several timeouted request before posting logs.

Have you checked out ltrace/strace?

No, I don't, because they output very huge, and I don't know what exactly I should looking there.