RE: [squid-users] RE: Two squid servers to talk to each other before the internet

Sorry the config file is a bit of a legacy of 5 people who been touching the config in the past years.

So...I've did what you said:
-removed "always_direct allow all"
-cahnged "via off" -> "via on"
-changed " icp_access deny all" -> " icp_access allow all"

I have a result as an empty white page in my browser, and nothing in the access.log
Still no UDP_*_ or any sign of the servers talking to the other, but worst because I dont even see myself accessing to any site anymore.
Might be something still missing?

Thanks,
Tibby

-----Original Message-----
From: Amos Jeffries [mailto:squid3_at_treenet.co.nz]
Sent: Wednesday, September 08, 2010 2:42 PM
To: squid-users_at_squid-cache.org
Subject: Re: [squid-users] RE: Two squid servers to talk to each other before the internet

On 08/09/10 21:20, Tóth Tibor Péter wrote:
> Hi Amos!
> Here is my config file:
>
> http_port 8080
> hierarchy_stoplist cgi-bin ?
> acl QUERY urlpath_regex cgi-bin \?
> cache deny QUERY

If you have a squid newer then 2.6.STABLE18 you can safely remove these
QUERY line. It will improve your hit rates a lot. The new
refresh_pattern below replaces them.

hierarchy_stoplist is still needed up to squid 3.1. After that it can go
too.

> acl apache rep_header Server ^Apache
>
> cache_peer ######THE_IP_OF_THE_SIBLING###### sibling 3128 3130
> #prefer_direct off
>
> cache_mem 1024 MB
> maximum_object_size 4096 KB
> minimum_object_size 0 KB
>
> cache_dir ufs /var/spool/squid3 75000 32 256
>
> error_directory /usr/share/squid3/errors/English
>
> logformat squid %ts.%03tu %6tr %>a %Ss/%03Hs %<st %rm %ru %un %Sh/%<A %mt
> access_log /var/log/squid3/access.log squid
> cache_store_log none
> logfile_rotate 1
>
> debug_options ALL,1
> cache_log syslog
>
> ftp_user ftp@
>
> hosts_file /etc/hosts
>
> refresh_pattern ^ftp: 1440 20% 10080
> refresh_pattern ^gopher: 1440 0% 1440

refresh+pattern -i (/cgi-bin/|\?) 0 0% 0

> refresh_pattern . 0 20% 4320
>
> httpd_suppress_version_string on
>
> acl manager proto cache_object
> acl localhost src 127.0.0.1/255.255.255.255

acl localhost src 127.0.0.1

> acl to_localhost dst 127.0.0.0/8

acl to_localhost dst 127.0.0.0/8 0.0.0.0/32

> acl SSL_ports port 443 # https
> acl Safe_ports port 21 # ftp
> acl Safe_ports port 80 # http
> acl Safe_ports port 880 # http
> acl Safe_ports port 443 # https
> acl Safe_ports port 1025-65535
> acl purge method PURGE
> acl CONNECT method CONNECT
>
> http_access allow manager localhost
> http_access deny manager
> http_access allow purge localhost
> http_access deny purge
> http_access deny !Safe_ports
> http_access deny CONNECT !SSL_ports
> http_access deny to_localhost
>
> acl desktop-clients src 0.0.0.0/0

acl desktop-clients src all

... that is not good either way. "all" and the above numbers mean the
entire Internet is one of your desktop-clients.

> acl denied-desktop-clients src ####SOME_IP_ADDRESS SOME_OTHER_IP AND_SOME_MORE_IP_ADDRESSES####
> acl denied-domains dstdom_regex -i "/etc/squid3/denied-hosts.acl"

If its just domain names and wildcard sub-domains on that list
"dstdomain" is faster than "dstdom_regex".

>
> http_access deny denied-desktop-clients
> http_access deny denied-domains
> http_access allow desktop-clients
> http_access allow localhost
> http_access deny all
>
> http_reply_access allow all
>

You will need to permit ICP access between the siblings or they will not
trade replies like you want. You will see a lot of UDP_*_MISS with icp
access denied.

> icp_access deny all

> htcp_clr_access deny all
>
> htcp_access deny all
> miss_access allow all
>
> visible_hostname ####THE_HOSTNAME.DOMAIN_OF_THIS_HOST####
> via off

via is REQUIRED to be ON when linking proxies together like this. It's
what prevents a single request looping around between your sibling
proxies until all existing network sockets are used up.

> forwarded_for off
>
> cachemgr_passwd ####SOME_PASSWORD#### all
> always_direct allow all

There is your main problem. "always_direct" FORCES the proxy to ignore
its sibling, not to even bother trying a lookup there.
Remove this and you will start to see requests between them.

Amos

>
> -----Original Message-----
> From: Amos Jeffries [mailto:squid3_at_treenet.co.nz]
> Sent: Tuesday, September 07, 2010 1:51 PM
> To: squid-users_at_squid-cache.org
> Subject: Re: [squid-users] RE: Two squid servers to talk to each other before the internet
>
> On 07/09/10 22:49, Tóth Tibor Péter wrote:
>>> Is there a way to check if squids are talking to each other?
>>
>> The access.log of each proxy will contain entries for messages going to
>> and from the sibling.
>>
>> On a basic setup like you have so far, expect to see SIBLING hit/miss
>> codes sometimes. UDP_SIBLING_* are the ICP messages flowing between the
>> siblings as they check whether the other has an object. TCP_SIBLING_HIT
>> are the actual HTTP reply objects being fetched.
>>
>> Amos
>>
>> Hi Amos!
>>
>> I dont see anything inside of the access.log about the SIBLING. Could it be the log level somewhere? Or something is configured wrong?
>> In fact I only see logs starting with TCP_(MISS/200, REFRESH_UNMODIFIED, IMS_HIT/304, MEM_HIT/200, TCP_HIT/200, etc...) Nothing with UDP_
>>
>> What can be wrong?
>>
>> Tibby
>
> What does your squid.conf look like then?
>
>
> Amos


--
Please be using
   Current Stable Squid 2.7.STABLE9 or 3.1.8
   Beta testers wanted for 3.2.0.2
Received on Thu Sep 09 2010 - 14:55:00 MDT