Re: [squid-users] delay pool configuration and query - PL PL Respond

> lokesh.khanna@accelonafrica.com wrote:
>
>>-----Original Message-----
>>From: Chris Robertson [mailto:crobertson@gci.net]
>>Sent: Tuesday, April 11, 2006 1:31 AM
>>To: squid-users@squid-cache.org
>>Subject: Re: [squid-users] delay pool configuration and query - PL PL Respond
>>
>>lokesh.khanna@accelonafrica.com wrote:
>>
>>
>>
>>>Hi
>>>
>>>I posted my question earlier also but no one replied and I am not
>>>getting enough information on squid website also. So pl reply to me.
>>>
>>>Below is my question and configuration
>>>
>>>I am using class 3 delay pool and objective is to allocate 128kbps to
>>>each host which belongs to 10.60.0.0/16 subnet.
>>>Below is my configuration
>>>
>>>
>>>acl only128kusers src 10.60.0.0/255.255.0.0
>>>delay_pools 1
>>>delay_class 1 3
>>>delay_access 1 allow only128kusers
>>>delay_access 1 deny all
>>>delay_parameters 1 -1/-1 -1/-1 16000/16000
>>>
>>>Questions :
>>>
>>>1. I noticed users get 128kbps bandwidth when they do download. But I
>>>also noticed when I do normal browsing, peak some time goes high up to
>>>600 kbps or 700 kbps. I used Packet shaper to monitor end user bandwidth
>>>
>>>
>>>usage. So I want to know what is the policy, why I am seeing peak up to
>>>600 or 700 kbps??
>>>
>>>
>>>
>>>
>>Where are you doing this monitoring? Between the Internet and Squid or
>>between Squid and the clients. Cached traffic is not affected by delay
>>pools.
>>
>>
>>
>>>2. I want to know if there is any command in squid to check how many
>>>users are falling in delay pool policy etc. this will help in doing any
>>>troubleshooting.
>>>
>>>
>>>
>>>
>>Either use the squidclient on the proxy to access the cache manager...
>>
>>/path/to/squidclient cache_object://localhost/menu
>>
>>...or set up a web server with the cachemgr.cgi and give it access to
>>your cache. Look for a menu object relating to delay pools. I don't
>>have a server set up with delay pools, so I can't be more specific.
>>
>>
>>
>>>Thanks - LK
>>>
>>>
>>>
>>>
>>Chris
>>
>>
>Thanks Chris
>
>I am seeing peaks between Squid and Clients. Is there any way to force
>cached data also to go thru delay pool?
>
>
You might be better served by a traffic shaper (such as tc on Linux or
dummynet pipes on FreeBSD). As far as I can tell, delay pools are just
designed to preserve traffic between Squid and the internet.

>What is delay_initial_bucket_level. If I make it 0 or make it 100 , what
>will happen? Pl give some example. I have read FAQ and other documents
>but I couldn't understand this.
>
>
Considering you have defined the same maximum and restore bucket size,
this parameter is meaningless. If you gave a larger maximum bucket
size, this would determine how full the maximum bucket would start with
a new client.

>Another thing which I want to know is - Is delay pool going to take more
>resources from server? Do I need to add more memory or CPU? I am
>planning to use class c delay pools. And I have nearly 3000 users.
>
>
unquestionably, more resources will be required (Squid is being required
to track more information). I can't possibly begin to tell you how much
though.

>Thanks - LK
>
>
>
>
Chris
Received on Tue Apr 11 2006 - 14:42:35 MDT