Nathan,
I used to run two squids on my servers which provide very similar, but
different services. I'm using SmpScale to simplify that configuration
into a single squid instance with 2 workers with slightly different
configurations. Honestly, there's one line difference, here's the top of
my squid.conf:
workers 2
if ${process_number} = 1
http_port 80
endif
if ${process_number} = 2
http_port 81
hosts_file /etc/squid3/hosts
endif
I realize that in the future the DNS cache may be shared between
workers, but for the moment it isn't so this setup works :)
So I could add something like this to make squid happy, and it won't
even open the port:
if ${process_number} = 3
http_port = 64000
endif
But that seems like a hack to me.
--Will
On 01/15/2014 06:14 PM, Nathan Hoad wrote:
> Hi Will,
>
> Why are you giving each worker a unique set of ports? Typically you
> configure one set of ports for all workers, and let the operating
> system handle the underlying machinery of sharing the ports across the
> worker processes.
>
> See here for more details:
> http://wiki.squid-cache.org/Features/SmpScale#Who_decides_which_worker_gets_the_request.3F
>
> Thanks,
>
> Nathan.
> --
> Nathan Hoad
> Software Developer
> www.getoffmalawn.com
>
>
> On Thu, Jan 16, 2014 at 9:19 AM, Will Roberts <ironwill42_at_gmail.com> wrote:
>> Hi,
>>
>> I'm working with an SmpScale configuration with 2 workers defined. Each
>> worker has its own set of unique ports that it listens on. The coordinator
>> process doesn't have any http_port lines and generates tons of these
>> warnings:
>>
>> ERROR: No forward-proxy ports configured
>>
>> That doesn't seem great, am I missing something? I'm sure I can work around
>> this by giving it some dummy port to listen on, but I'd rather not if it
>> doesn't really need it.
>>
>> Thanks,
>> --Will
Received on Wed Jan 15 2014 - 23:22:03 MST
This archive was generated by hypermail 2.2.0 : Thu Jan 16 2014 - 12:00:05 MST