lightsquid report filtering
Hello everyone, continuing with my project. I managed to configure squid as non-transparent (as proof). It works perfectly!
I installed lightsquid and SARG to evaluate the reports generated. I get something similar to this:
** r2---sn-x1x7zn7s.c.2mdn.net:443 1 864 017 64.2 M 1.0%
12 static.clarin.com:443 7 859 585 65.1 M 1.0%
13 arc-anglerfish-arc2-prod-infobae.s3.amazonaws.com:443 1 728 070 65.8 M 0.8%
14 d3nn82uaxijpm6.cloudfront.net:443 6 573 861 66.3 M 0.6%
15 www.cadena3.com.ar 3 547 366 66.8 M 0.6%
16 securepubads.g.doubleclick.net:443 13 533 625 67.3 M 0.6%
17 cdn.lavoz.com.ar:443 12 516 833 67.8 M 0.6%
18 use.fontawesome.com:443 4 498 180 68.3 M 0.5%
The report is for management areas, which are interested in accessed sites, and browsing time. So you would need to report the accessed site and filter URL with advertising and other cache
Has anyone succeeded in putting together something similar?
KOM last edited by KOM
I think Lightsquid does all of that. Sarg is no longer a package and hasn't been for some time so I have no comment on that. IIRC Lightsquid will tell you bytes transferred and time of day it was requested for each hour, but I don't think it tells you exactly how long they were on a site because that's hard if not impossible to figure out.
@KOM Is it possible in LightSquid, in the Squid User Access Report to display a logged in user name from the captive portal, via the user logging in with their account created under the user manager?
I get an IP address for user, and then a "?" for real name. Is there a way to populate the user field with the actual logged in username and not the IP address?
@KOM thanks for answering!
Ok I understand that it is difficult to know exactly how long a user has a focus on a browser or specific website.
Lightsquid already works for me, is there any way to filter that report? For example, just show sites that start with * www. *. * ??
I think that would be very useful with that
KOM last edited by
@WD_Doug If none of the stock methods for reporting the user work for that, then you're going to have to hack something up. These might be useful:
@mauricio2669 Not really, and even the report itself is built on the fly from several pieces so it's not easy to parse yourself.