[squid-users] Re: Delay pools bucket refill

From: Johannes Buchner <buchner.johannes_at_gmx.at>
Date: Thu, 25 Dec 2008 18:56:59 +0100

On Tue, 23 Dec 2008 11:01:33 -0800 (PST)
Chuck Kollars <ckollars9_at_yahoo.com> wrote:

> > ... If I make a time-based acl with a delay-pool, does it refill
> > in the time the acl is inactive or is the amount "stopped" and
> > continued when the acl starts again?
>
> It doesn't matter hardly at all. The bucket will "overflow" and never
> grow beyond the second parameter no matter what. So at most you're
> just asking if the bucket _starts_out_ "full" or "empty" when the ACL
> starts again. After a few tens of seconds the initial value won't
> make any difference; you're just talking about a transient condition
> that might last up to one minute.

You skipped the part where I tell the parameters I'd like to use:
>> Like, if I have a pool acl going from 9:00 till 20:00 with a size of
>> 3GB and a rate of 1200 B/s, and a client runs low on the bucket at
>> 20:00. What will he be able to download at 9:00 the next day?
It will take one month to refill the bucket. That is what I want to
do: offer a 3GB download limit each month, and if the bucket is empty,
the user will be able to download at 1200B/s (or has to wait a while).

It matters in my scenario.

On Wed, 24 Dec 2008 15:47:12 +1300
Amos Jeffries <squid3_at_treenet.co.nz> wrote:

> Johannes Buchner wrote:
> > Hi!
> >
> > I have a question about delay_pools: If I make a time-based acl
> > with a delay-pool, does it refill in the time the acl is inactive
> > or is the amount "stopped" and continued when the acl starts again?
>
> Pools refill at the constant rate unless the are full or
> reconfigured. Client usage is not taken into consideration on the
> filling, only on the emptying.
I'm not talking about client usage, just wether the acl is active or
not (since it has time constraints).

> > ...if I would define one bucket for 9:00 till 20:00 and another
> > one for 20:00 till 9:00 of different sizes and rates, would they
> > share their amount?
>
> There's really only one bucket per node at a time no matter what. (It
> may be possible with some uses of ACLs to make the first bucket "go
> away" and the second bucket [exactly like it] "replace" it. In that
> case I'd reframe your question as 'does the existing content of the
> first bucket become the initial value in the second bucket?'.) Again,
> it doesn't much matter. Since every bucket will "spill over" when its
> defined size is reached, you're again just asking what will happen in
> the first few seconds, whether the bucket will initially be "empty"
> or "same as previous" or "full". After a few tens of seconds any such
> initial value will be completely swamped out by the ongoing action of
> the system.

Amos has a different answer on this one:

On Wed, 24 Dec 2008 15:47:12 +1300
Amos Jeffries <squid3_at_treenet.co.nz> wrote:
> Correct. No. They are different pools.

Amos, ckollars, thank you for your answers.

I'll just try it out and report back.

Regards,
Johannes

-- 
Emails knnen gendert, geflscht und eingesehen werden. Signiere oder
verschssele deine Mails mit GPG.
http://web.student.tuwien.ac.at/~e0625457/pgp.html

Received on Thu Dec 25 2008 - 17:57:24 MST

This archive was generated by hypermail 2.2.0 : Fri Dec 26 2008 - 12:00:01 MST