httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jim Jagielski <...@jaguNET.com>
Subject Re: Negative mod_proxy ping_timeout..
Date Tue, 09 Feb 2016 17:37:07 GMT

> On Feb 9, 2016, at 11:49 AM, Yann Ylavic <ylavic.dev@gmail.com> wrote:
> 
> On Tue, Feb 9, 2016 at 5:42 PM, Jim Jagielski <jim@jagunet.com> wrote:
>> Why the change to modules/proxy/mod_proxy_hcheck.c?
> 
> The call to ap_proxy_connect_backend() just above already calls
> ap_proxy_is_socket_connected() for any reused connection, so why check
> it twice?
> 

I recall a discussion somewhat long ago where we also
do a specific check after paths where we "just checked"
and the reasoning was to avoid race conditions, esp in
situations where we created but then it gets immediately killed.

I am sure there are other places where we do the same,
likely w/ comments. :)

> If the connection is not reused but a new one created instead, the
> check is also pointless IMHO.


Mime
View raw message