Re: ICP not working as expected with Squid 2.0.PATCH2

From: Janos Farkas <Janos.Farkas-lists+priv-#dQC2uAaJDX7*>
Date: Wed, 14 Oct 1998 14:37:52 +0200

On 1998-10-14 at 13:07:31, John Line wrote:
> With cache_peer lines ending " parent 8080 3130", ICP requests do not
> work as expected (though the same config works with 1.NOVM.22) and instead
> (with some extra logging enabled) I see e.g.
> 1998/10/14 12:38:46| neighborsUdpAck: Unexpected ICP_MISS for FC63DC7ADFE328FA18F859B1A47C1974

Quite a few of use has been bitten by it, check dead_peer_timeout in
squid.conf. It's defaulting to 10 seconds, meaning if your squid hasn't
heard of a peer in 10 seconds, it won't use it. For us with very light
loads, i.e. very few users, so 10 seconds without a query is quite a
common, it means squid will always try to go direct for the first query
after a longer pause. You might want to up it for maybe a hour, or so,
it worked wonders for me.. Still, I don't quite like it this way :)

Janos - Don't worry, my address is real.  I'm just bored of spam.
Received on Wed Oct 14 1998 - 06:39:58 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:42:29 MST