nginx -s reload does not work with proxy directives

[Maxim Dounin]

Hello!

On Tue, May 31, 2022 at 06:01:01PM +0200, evald ibrahimi wrote:

> Just encounter this issue on the latest version of nginx 1.22.0.
> Basically i added the "proxy_read_timeout 600s;" in a server directive and
> run "nginx -s reload".
> I was getting still "upstream timed out (110: Connection timed out) while
> reading response header from upstream, client: " after 60 seconds which is
> the default value.
> Than i said, let's restart the whole service with systemctl restart nginx
> and it worked..no more upstream timeout.

This suggests that reload wasn't successful for some reason.

There are various legitimate cases when reload might fail even if 
the configuration is perfectly correct: for example, if there are 
conflicting settings for a shared memory zone (e.g., you cannot 
change size of a shared memory zone), or when using conflicting 
listening sockets on Linux (such as "listen 80;" and "listen 
127.0.0.1:80 bind;").  Detailed information about why reload 
failed is expected to appear in the error log.

Further, bugs in 3rd party modules might additionally break 
configuration reloads.  If there are any, you may want to test if 
things work correctly without 3rd party modules.

Hope this helps.

-- 
Maxim Dounin
http://mdounin.ru/