Seems to have done the trick, also needed to increase the number of
file descriptors to handle the extra helpers.
At peak load now using 33 (30 was our Default), what i can only assume
has something to do with problems related to queuing theory and
exponential wait times occurred due to this.
Hopefully this helps someone in the future.
-Jason.
On Mon, Jul 9, 2012 at 10:37 PM, Jason Leschnik <lesch...@gmail.com> wrote:
> #!/bin/bash
>
> REFRESH=5
>
> while [ 1 ]
> do
> /usr/sbin/squidclient mgr:ntlmauthenticator | tail -n +20 | head -n 39
> sleep ${REFRESH}
> clear
> done
>
> I quickly knocked up this so i can monitor it tomorrow at work :D thanks
>
> On Mon, Jul 9, 2012 at 10:02 PM, Amos Jeffries <squ...@treenet.co.nz> wrote:
>> On 9/07/2012 11:50 p.m., Jason Leschnik wrote:
>>>
>>> Thanks for the reply :)
>>>
>>> I did some basic math on the cache.log and came up with about 80
>>> helpers we need. I will monitor both the cache.log + the general user
>>> experience to see if this improves the situation.
>>
>>
>> In the cache manager repots there is a "ntlmauthenticators" report which
>> lists all the helpers and how loaded they are. For efficiency and easy
>> management the helpers are loaded so the first ones get a lot of requests
>> then it tails off towrads the last one. You want to get it so that you have
>> enough to cover peak traffic, but the last of your set is rarely used.
>>
>> Amos
>>
>
>
>
> --
> Regards,
> Jason Leschnik.
>
> [m] 0432 35 4224
> [w@] jason dot leschnik <at> ansto dot gov dot au
> [U@] jml...@uow.edu.au
--
Regards,
Jason Leschnik.
[m] 0432 35 4224
[w@] jason dot leschnik <at> ansto dot gov dot au
[U@] jml...@uow.edu.au
