On 27.03.2012 10:13, Carlos Manuel Trepeu Pupo wrote:
On Sat, Mar 24, 2012 at 6:31 PM, Amos Jeffries <squid3@xxxxxxxxxxxxx>
wrote:
On 25/03/2012 7:23 a.m., Carlos Manuel Trepeu Pupo wrote:
On Thu, Mar 22, 2012 at 10:00 PM, Amos Jeffries wrote:
On 23/03/2012 5:42 a.m., Carlos Manuel Trepeu Pupo wrote:
I need to block each user to make just one connection to download
specific extension files, but I dont know how to tell that can
make
just one connection to each file and not just one connection to
every
file with this extension.
i.e:
www.google.com #All connection that required
www.any.domain.com/my_file.rar #just one connection to that file
www.other.domain.net/other_file.iso #just connection to this file
www.other_domain1.com/other_file1.rar #just one connection to
that file
I hope you understand me and can help me, I have my boss hurrying
me !!!
There is no easy way to test this in Squid.
You need an external_acl_type helper which gets given the URI and
decides
whether it is permitted or not. That decision can be made by
querying
Squid
cache manager for the list of active_requests and seeing if the
URL
appears
more than once.
Hello Amos, following your instructions I make this
external_acl_type
helper:
#!/bin/bash
result=`squidclient -h 192.168.19.19 mgr:active_requests | grep -c
"$1"`
if [ $result -eq 0 ]
then
echo 'OK'
else
echo 'ERR'
fi
# If I have the same URI then I denied. I make a few test and it
work
for me. The problem is when I add the rule to the squid. I make
this:
acl extensions url_regex "/etc/squid3/extensions"
external_acl_type one_conn %URI /home/carlos/script
acl limit external one_conn
# where extensions have:
\.(iso|avi|wav|mp3|mp4|mpeg|swf|flv|mpg|wma|ogg|wmv|asx|asf|deb|rpm|exe|zip|tar|tgz|rar|ppt|doc|tiff|pdf)$
http_access deny extensions limit
So when I make squid3 -k reconfigure the squid stop working
What can be happening ???
* The helper needs to be running in a constant loop.
You can find an example
http://bazaar.launchpad.net/~squid/squid/3.2/view/head:/helpers/url_rewrite/fake/url_fake_rewrite.sh
although that is re-writer and you do need to keep the OK/ERR for
external
ACL.
Sorry, this is my first helper, I do not understand the meaning of
running in a constant loop, in the example I see something like I do.
Making some test I found that without this line :
result=`squidclient -h 192.168.19.19 mgr:active_requests | grep -c
"$1"`
the helper not crash, dont work event too, but do not crash, so i
consider this is in some way the problem.
Squid starts helpers then uses the STDIN channel to pass it a series of
requests, reading STDOUt channel for the results. The helper once
started is expected to continue until a EOL/close/terminate signal is
received on its STDIN.
Your helper is exiting without being asked to be Squid after only one
request. That is logged by Squid as a "crash".
* "eq 0" - there should always be 1 request matching the URL. Which
is the
request you are testing to see if its >1 or not. You are wanting to
deny for
the case where there are *2* requests in existence.
This is true, but the way I saw was: "If the URL do not exist, so
can't be duplicate", I think isn't wrong !!
It can't not exist. Squid is already servicing the request you are
testing about.
Like this:
receive HTTP request -> (count=1)
- test ACL (count=1 -> OK)
- done (count=0)
receive a HTTP request (count-=1)
- test ACL (count=1 -> OK)
receive b HTTP request (count=2)
- test ACL (count=2 -> ERR)
- reject b (count=1)
done a (count=0)
* ensure you have manager requests form localhost not going through
the ACL
test.
I was making this wrong, the localhost was going through the ACL, but
I just changed !!! The problem persist, What can I do ???
which problem?
Amos
