Re: TCP_REFRESH_HIT a MISS

From: Maciej Kozinski <[email protected]>
Date: Wed, 4 Aug 1999 10:30:26 +0200 (CEST)

Hi Jens,

Jens-S. Voeckler:
>
> 3) For the developers: Is it possible to express the parent/sibling
> relationship configured in squid.conf appropriately in the log file for
> cache digest hits, e.g. log a CD_(PARENT|PEER)_HIT?
>

 Could you write more about why do you need that? It would complicate log file
processing ;) I don;'t have an idea how could it be usable.
 I wonder if it really does make a sense... Making this kind of difference sure
does make a sense in case of ICP, when you ask for content information, but also
about route availability information (FIRST_PARENT_MISS?). When you have cache
digests, in fact it doesn't matter that the hit comes from parent or sibling -
every hit is sibling-like (in the meaning of ICP), because using CD you're
interested only in cache having desired content, not these which do not have it.
The cache behaviour when no CD hit sould be like for default parent or round
robin (when more than one parent). But I think the parents will die soon... in
complicated network topology their existence does make a sense too.

Regards,
Maciej

-- 
           Maciej Kozinski         http://www.uck.uni.torun.pl/~maciek/ 
	   Remember: Un*x _IS_ user friendly... It's just selective about
		     who it's friends are.
Received on Wed Aug 04 1999 - 02:17:36 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:47:50 MST