3 messages in ru.sysoev.nginxRe: monitoring for load balancing
FromSent OnAttachments
rkmr...@public.gmane.orgNov 27, 2007 3:29 pm 
Ben ReubensteinNov 27, 2007 6:26 pm 
Grzegorz NosekNov 29, 2007 3:51 am 
Actions with this message:
Paste this link in email or IM:
Paste this link in email or IM:
Atom feed for this thread
Paste this URL into your reader:
Subject:Re: monitoring for load balancingActions...
From:Grzegorz Nosek (grze@public.gmane.org)
Date:Nov 29, 2007 3:51:05 am
List:ru.sysoev.nginx

Hi Ben,

2007/11/28, Ben Reubenstein <benr-gpfhg/Mf3@public.gmane.org>:

If I understand the current proxying system correctly, it is just a simple round robin. So when a request comes in the requests are handed out from the first process and so on. This can cause a request

Yep, this is correct (though the exact round-robin method apparently changed between 0.5 and 0.6, with regard to peers having weight != 1).

to be "stuck" in line. There is a fair scheduler in the works, more info here:

http://brainspl.at/articles/2007/11/09/a-fair-proxy-balancer-for-nginx-and-mongrel

As far as monitoring the individual processes, I will often just pop into the box and take a look via various *nix tools.

upstream_fair does not provide any statistics to the outside world currently, but I want to implement some, once any bugs are ironed out.

Best regards, Grzegorz Nosek