Re: http rfc pointer to web-server-overload logic

From: Henrik Nordstrom <hno@dont-contact.us>
Date: Thu, 12 Feb 1998 23:30:27 +0100

Lincoln Dale wrote:
,,
> does anyone have a pointer to where abouts and in what HTTP RFC
> it states that servers can simply refuse connections.
>
> i've gone through RFC 2068... with no success.

RFC 2068, section 10.5.4 (503 Service Unavailable)

     Note: The existence of the 503 status code does not imply that a
     server must use it when becoming overloaded. Some servers may wish
     to simply refuse the connection.

But I can't find anything on how to handle clusters (more than one
server for one URL). The only place it is mentioned is in a short note
on DNS spoofing. I can't find anything about restarting failed
connections (either time out, refused or status 503)...

But common sense gives that it is a good idea. If we know that a web
server is driven by a cluster (it has several IP addresses), then if one
of these fails I beleive common sense is to try another.

---
Henrik Nordström
Sparetime Squid Hacker
Received on Thu Feb 12 1998 - 15:21:39 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:38:52 MST