Re: [squid-users] cache peering - load balancing and failover

From: Henrik Nordstrom <[email protected]>
Date: Thu, 15 Nov 2007 02:22:13 +0100

On ons, 2007-11-14 at 19:09 +0530, Manu Garg wrote:
> Here is my problem:
>
> I have a cache server at location X: cache.X. This server peers up
> with cache servers at location Y and Z:
> cache1.Y,
> cache2.Y,
> cache1.Z,
> cache2.Z.
>
> I want cache.X to talk to cache[12].Y in round robin manner as long as
> they are accessible. Peering should failover to cache[12].Z only if
> X->Y connectivity is down.
>
> Is it possible to do this with squid? Probably, something of the sort -
>
> cache_peer cache1.Y parent 3128 0 no-query round-robin priority=1
> cache_peer cache2.Y parent 3128 0 no-query round-robin priority=1
> cache_peer cache1.Z parent 3128 0 no-query round-robin priority=2
> cache_peer cache2.Z parent 3128 0 no-query round-robin priority=2

Looks fine to me, but add a relatively short connect timeout to the mix
for quicker failover.

When Squid has failed to connect to a peer 10 times in a row it's
declared dead and is taken out until it starts responding again.

Regards
Henrik

Received on Wed Nov 14 2007 - 18:22:19 MST

This archive was generated by hypermail pre-2.1.9 : Sat Dec 01 2007 - 12:00:02 MST