[squid-users] Skype for Business behind a transparent squid (TProxy) HTTP/S

Tue Dec 6 18:27:13 UTC 2016

Now you can enhance the script by adding manually the ntop skype related networks based on:
https://github.com/ntop/nDPI/blob/d9a2d9a6bd4d476d666d26cb713952760a975d92/src/lib/ndpi_content_match.c.inc#L286

/*
Skype (Microsoft CDN)
157.56.135.64/26, 157.56.185.0/26, 157.56.52.0/26,
157.56.53.128/25, 157.56.198.0/26
157.60.0.0/16, 157.54.0.0/15 
13.107.3.128/32
13.107.3.129/32
111.221.64.0 - 111.221.127.255
91.190.216.0/21 (AS198015 Skype Communications Sarl)
91.190.218.0/24
40.126.129.109/32
65.55.223.0/26
*/

If you need help scripting this let me know.

Eliezer

----
http://ngtech.co.il/lmgtfy/
Linux System Administrator
Mobile: +972-5-28704261
Email: eliezer at ngtech.co.il

From: Sameh Onaissi [mailto:sameh.onaissi at solcv.com] 
Sent: Tuesday, December 6, 2016 7:29 PM
To: Eliezer Croitoru <eliezer at ngtech.co.il>
Cc: Amos Jeffries <squid3 at treenet.co.nz>; squid-users at lists.squid-cache.org
Subject: Re: [squid-users] Skype for Business behind a transparent squid (TProxy) HTTP/S

Hello, 

OK, I added the ssl_bump slice on the skype domains text file
I installed ipset and ran the script.

Now access.log has much less skype related logs:

What is left is:
1481044996.398   3412 10.0.0.11 TAG_NONE/200 0 CONNECT 132.245.1.32:443 - ORIGINAL_DST/132.245.1.32 -
1481044996.423      0 10.0.0.11 TAG_NONE/400 3998 REGISTER sip:solcv.comSIP/2.0 - HIER_NONE/- text/html
1481045000.296    372 10.0.0.11 TAG_NONE/200 0 CONNECT 134.170.113.207:443 - ORIGINAL_DST/134.170.113.207 -
1481045000.325      0 10.0.0.11 TAG_NONE/400 3998 REGISTER sip:solcv.comSIP/2.0 - HIER_NONE/- text/html
1481045008.685   4259 10.0.0.11 TAG_NONE/200 0 CONNECT 134.170.113.207:443 - ORIGINAL_DST/134.170.113.207 -
1481045008.726      0 10.0.0.11 TAG_NONE/400 3998 REGISTER sip:solcv.comSIP/2.0 - HIER_NONE/- text/html

although http://solve.com is in the text file.

I ran whois on the first IP and got:

NetRange:       132.245.0.0 - 132.245.255.255
CIDR:           132.245.0.0/16
NetName:        MICROSOFT

Same with the 134.170. address. Can we slice that range?

Sameh Onaissi
Ingeniero de Soporte
Sol Cable Visión
Cel: 316-3023424
Email: mailto:sameh.onaissi at solcv.com

Piensa en el medio ambiente antes de imprimir este email. 

On Dec 6, 2016, at 12:11 PM, Eliezer Croitoru <mailto:eliezer at ngtech.co.il> wrote:

Hey,

Depends on your OS you will need to installthe  ipset package.
Try to run “apt-get install ipset”.
And then run the script.

Eliezer

----
http://ngtech.co.il/lmgtfy/
Linux System Administrator
Mobile: +972-5-28704261
Email: mailto:eliezer at ngtech.co.il
<Untitled Attachment 1.jpg>

From: Sameh Onaissi [mailto:sameh.onaissi at solcv.com] 
Sent: Tuesday, December 6, 2016 5:23 PM
To: Amos Jeffries <mailto:squid3 at treenet.co.nz>
Cc: Eliezer Croitoru <mailto:eliezer at ngtech.co.il>
Subject: Re: [squid-users] Skype for Business behind a transparent squid (TProxy) HTTP/S

Amos, thanks for the reply. 

This is getting more confusing.

I changed the script to: http://pastebin.com/jLgywstg

And I ran it, but I am getting errors:

sudo sh http://bypass.sh/ + iptables -t mangle -L PREROUTING + grep bypasspool + [ 1 -ne 0 ] + iptables -t mangle -I PREROUTING -m set --match-set bypasspool dst,src -j DIVERT iptables http://v1.6.0/ Set bypasspool doesn't exist. Try `iptables -h' or 'iptables --help' for more information. + ipset create bypasspool hash:ip http://bypass.sh/ 10: http://bypass.sh/ ipset: not found + read item + echohttp://lyncdiscover.solcv.com/ http://lyncdiscover.solcv.com/ + host -4 http://lyncdiscover.solcv.com/ + grep has address + awk {print $4} + xargs -l1 ipset add bypasspool xargs: ipset: No such file or directory + read item + echo http://webdir0a.online.lync.com/http://webdir0a.online.lync.com/ + host -4 http://webdir0a.online.lync.com/ + grep has address + awk {print $4} + xargs -l1 ipset add bypasspool xargs: ipset: No such file or directory

… this goes on the same for all the domains in the text file

My iptables is still <http://pastebin.com/SqpbmYQQ>

I did not quite understand what you meant by 
You should test whether -m set or -m socket work faster and put that one
first. My change above places it at line 2 (after -m socket) assuming
your iptables script is still <http://pastebin.com/SqpbmYQQ>

should I incorporate the bypass script into my iptables.sh script? run iptables first then bypass?

On a side note, would adding ssl_bump exceptions to squid.conf do it?
Something like: 

acl skype_domains <path to file>
ssl_bump splice skype_domains
ssl_bump bump all

Again, thanks again for your help.

<Untitled Attachment 2.jpg> Piensa en el medio ambiente antes de imprimir este email.

On Dec 6, 2016, at 9:50 AM, Amos Jeffries <mailto:squid3 at treenet.co.nz> wrote:

On 7/12/2016 3:19 a.m., Sameh Onaissi wrote:
Hello,

I tried doing the changes to nat/REDIRECT in iptables.sh and I must have messed up somewhere, so I am sticking with mangle/tproxy for now since squid is working with them.

How can I change Eliezer’s script to mangle/tproxy? https://gist.github.com/elico/e0faadf0cc63942c5aaade808a87deef

Excuse my novice knowledge in iptables.

No worries.

You need to change where iptables attaches the 'bypasspool'. Both the
table/location (-t) and the jump/action (-j).

 iptables -t mangle -L PREROUTING |grep bypasspool
 if [ "$?" -ne "0" ];then

   iptables -t mangle -I 2 PREROUTING \
     -m set --match-set bypasspool dst,src \
     -j DIVERT

 fi

You should test whether -m set or -m socket work faster and put that one
first. My change above places it at line 2 (after -m socket) assuming
your iptables script is still <http://pastebin.com/SqpbmYQQ>

(Your script should do that line adding, not Eliezers - so that you can
be sure the order is always correct).

BTW: you should use iptables-save / iptables-restore instead of a slow
script calling iptables "manually". Those other tools will ensure there
are no gaps in the firewall initialization for nasty traffic to sneak
through.

I am looking at access.log to collect all domains I see heading to skype for business, as well as IPs. My question is, can I add the domains AND IPs into the domains-to-bypass.txt that the above script uses?

IIRC you should be able to use domain as the parameter to ipset. But it
will resolve the domain immediately and only add those IPs that it finds
at that time into the pool. Any future changes, or a hidden set of IPs
that rotate in/out will not be listed.

Amos