On Fri, Dec 29, 2017 at 02:52:39PM +0100, Lukas Tribus wrote:
> > For me it happens only when I have "option httpclose" in the configuration,
> > ie we end up in tunnel mode. I can't reproduce it with either keep-alive,
> > http-server-close nor forceclose. At least abortonclose is now safe
> > regarding this. Does this match your observations as well ?
>
> No, I'm using the default http-keep-alive mode.
Then I'm still missing something. At least it's nice because it means
there is definitely something odd that remains to be fixed!
> Haproxy calls shutdown() after the HTTP payload was transmitted, nginx
> in the default configuration or nc for that matter closes the
> connection (we see recvfrom = 0) and then we close():
I can't reproduce this one for now. I'm pretty sure that it's timing
related.
> 14:39:57.382142 sendto(9, "POST /111 HTTP/1.1\r\nuser-agent: "...,
> 169, MSG_DONTWAIT|MSG_NOSIGNAL, NULL, 0) = 169
(...)
> 14:39:57.392212 read(8, "\27\3\3\0(", 5) = 5
> 14:39:57.392458 read(8,
> "tQ\325'wD\2375\222\202\1\241\277\35\347\213;\221&\211\303g\322\226[\334\10Z\20\332\36s"...,
> 40) = 40
(...)
> 14:39:57.392868 sendto(9, "bla=bla", 7, MSG_DONTWAIT|MSG_NOSIGNAL, NULL, 0) =
> 7
> 14:39:57.393178 shutdown(9, SHUT_WR) = 0
(...)
It's interesting to see that the request was sent in two parts, which doesn't
happen in my case. That may be one of the differences. I'll try to play along
this.
thanks for the trace!
Willy
