Re: [squid-users] peer selection with weight=N

H wrote:
>> fre 2010-02-19 klockan 03:08 -0200 skrev H:
>>
>>> Thing is, all three as parent with no-query round-robin get equal load as
>>> supposed, but, giving one (any) of them weight=2 makes no difference, still
>>> gets the same load.
>> How much traffic do you have?
>
> 10Mbit/s plus 8Mbit/s on each parent cache
>
>
>>> So I thought doing this
>>>
>>> cache_peer parent_IP parent tport uport no-query [weight=2]
>>> cache_peer parent_IP parent tport uport no-query round-robin
>>> cache_peer parent_IP parent tport uport no-query round-robin
>> The peers all need to be configured equal except for the weight.
>>
>> For non-ICP/HTCP peers only round-robin has a definition of weight. The
>> other peering types all select just one or a ordered list..
>
>
> well, it was a try since weight does not weight ... btw, the above problem
> (not weighting) applies to icp as well
>

Well, Henrik omitted this...

When enabled, ICMP RTT measurement is used and weight=N is combined with
the RTT to bend the "closest" peer selection method and
weighted-round-robin method, based on an RTT-biased weighting.

But that all depends on netdb and ICMP being enabled to work reliably to
measure RTT.

>
>> Correct. Siblings are only sharing cache, not uplink. What differs a
>> sibling from a parent.
>>
>
> ok, but then this
>
> # use 'allow-miss' to disable Squid's use of only-if-cached when forwarding
> requests to siblings.
>
> eventually needs better explanation/clarification

Such as the current one?
   http://www.squid-cache.org/Doc/config/cache_peer/

>
> and the returned erro needs revision because it comes up ERR_ACCESS_DENIED and
> should be better ERR_ONLY_IF_CACHED_MISS

In which log? caller proxy or called peer?

Amos

-- 
Please be using
   Current Stable Squid 2.7.STABLE8 or 3.0.STABLE24
   Current Beta Squid 3.1.0.16