fastcgi retry upstream (transparent php-fpm restart)?

[Etienne Champetier]

Hi,

I'm using nginx with 1 php-fpm (via unix socket), and i'm wondering if
there is a way to make nginx retry on error, something like
fastcgi_next_upstream but for the same upstream.

I would like to be able to transparently restart php-fpm and not send some
502 to the clients

So, is there a way to make nginx retry the same upstream for X seconds (or
retry every Y seconds for max Z attempts or ...) (on errors like "connect()
failed (111: Connection refused) while connecting to upstream") ?

I'm open to any tricks

Thanks in advance
Etienne

p.s: I already know about atomic/transparent php deployment with
"$realpath_root" and multiple docroot (
https://github.com/zendtech/ZendOptimizerPlus/issues/126#issuecomment-24020445
)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nginx.org/pipermail/nginx/attachments/20150708/948859c1/attachment-0001.html>