Re: negative byte hit ratio...again

From: Vesselin Atanasov <>
Date: Mon, 18 Oct 1999 20:35:33 +0000 (UTC)


This could very easily happen if one of your cache clients connects
through squid to a very fast server. Then squid reads data stream from
server and this data stream takes all your bandwidth. If the client that
connects to squid has slow connection (e.g. 33.6 dialup), he cannot accept
data fast enough from squid and you get negative hit ration. The solution
that worked for me was to use delay pools and limit speed with which data
is fetched for dialup IPs to 33.6.

vesselin atanasov

On Wed, 13 Oct 1999, Jon Mansey wrote:

> Apologies for this which is probably an FAQ, but I did search the
> list archives and didnt find a useful answer.
> Can someone explain the meaning of a negative byte hit ratio?
> Cache information for squid:
> Request Hit Ratios: 5min: 40.5%, 60min: 32.9%
> Byte Hit Ratios: 5min: -2.7%, 60min: -12.7%
> Here it is only -2.7% but I have seen it swinging wildly from -120%
> (?) to +40% today.
> Can someone offer up a definition, or better a formula for how this
> value is calculated and what it means when it is negative?
> As I understand it, byte hit ratio = server-bytes-in / client-bytes-out
> Thus if I served 200MB to my clients, and of that the cache had to
> pull 100MB from origin servers, the byte hit ratio would be 50%.
> Is this correct?
> If the cache is pulling more on the server side than it is pushing to
> the clients, would it be negative then?
> What could cause that to happen? Many aborted or broken downloads?
> I am using the defaults for quick_abort on Version 2.2.STABLE5.
> TIA,
> Jon.

- ---------------------------------------
To get my public PGP key send an e-mail

Subject: get

Version: PGPfreeware 5.0i for non-commercial use
Charset: noconv

Received on Mon Oct 18 1999 - 11:39:24 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:48:57 MST