Re: [squid-users] 3.2.0.13: FATAL: dying from an unhandled exception: theGroupBSize > 0

On 24/10/11 20:49, David Touzeau wrote:
> Le lundi 24 octobre 2011 à 11:49 +1300, Amos Jeffries a écrit :
>> On 23/10/11 23:49, David Touzeau wrote:
>>> Dear
>>>
>>> I'm using the 3.2.0.13-20111022-r11381
>>>
>>> When set
>>>
>>> worker 2
>>>
>>> In squid.conf
>>>
>>> Has i have 2 CPUs
>>>
>>> Squid did not want to start with the following error
>>>
>>> terminate called after throwing an instance of 'TextException'
>>> what(): theGroupBSize> 0
>>> Aborted
>>
>> http://bugs.squid-cache.org/show_bug.cgi?id=3383
>>
>>>
>>> When you have a dedicated server for squid with multiple CPU, is it
>>> mandatory to set worker has the total CPUs installed on the computer ?
>>>
>>
>> No. It is completely optional.
>>
>> You may have more, or less. More is probably not very useful though, as
>> each worker still has the old Squid behaviour of scaling itself up to
>> consume the max cycles of a single CPU to meet traffic load.
>>
>> workers and SMP support in Squid-3.2 is just a way to simplify the
>> multi-instance configuration and management.
>>
>> Amos
>
>
> You means "a way to simplify the multi-instance configuration"
> So each Squid instance must run on a dedicated IP+port ?
>

Nope. Multiple instances with: shared config, shared listening ports,
shared memory cache, aggregated manager reports, aggregated SNMP responses.

Unless you configure otherwise using the ${process_number} macros.

Amos

-- 
Please be using
   Current Stable Squid 2.7.STABLE9 or 3.1.16
   Beta testers wanted for 3.2.0.13