[clug] force Apache to close connection on error

[Bob Edwards]

On 21/09/15 15:05, James Ring wrote:
> Maybe something like
> https://github.com/yaoweibin/nginx_upstream_check_module ? I've never
> used it so I can't vouch for it :)
>

That looks cool - I wonder why neither it, nor the much older
healthcheck_nginx_upstreams module it is derived from, have been
included in any of the Debian/Ubuntu packages for nginx (including
backports etc.)?

I'll give it a test and see what happens.

cheers,

Bob Edwards.

> On Sun, Sep 20, 2015 at 9:53 PM, Bob Edwards <Robert.Edwards at anu.edu.au> wrote:
>> Hi all,
>>
>> I am running a site with Nginx in front of Apache (actually in front
>> of lots of things, but am interested in Apache at the moment).
>>
>> The FOSS version of Nginx can do rudimentary fail-over if an "upstream"
>> (back-end) server is not responding.
>>
>> The commercial version of Nginx (Nginx Plus) can do fancier tricks like
>> detecting if the upstream server is responding with error code 500.
>>
>> Alas, if my site is "half-dead", Apache responds with error code 500
>> and my FOSS version of Nginx determines that as a valid response and so
>> keeps on sending requests to it.
>>
>> What I would like to do is get Apache to close the connection instead
>> of responding with error code 500. This probably breaks all kinds of
>> HTTP standards, but should suffice to trigger my Nginx to fail-over
>> to the backup server.
>>
>> Any ideas?
>>
>> cheers,
>>
>> Bob Edwards.
>>
>> --
>> linux mailing list
>> linux at lists.samba.org
>> https://lists.samba.org/mailman/listinfo/linux