color support, pcre and pf_ring issues

elof2 at sentor.se elof2 at sentor.se
Mon Oct 22 11:45:08 EDT 2012


> To turn color on, there is a new .rarc variable, and it takes yes and no for an answer.
>   RA_COLOR_SUPPORT="yes"

I would request an "auto" mode for RA_COLOR_SUPPORT, so ra* commands run 
via cron etc will yield plain text while ra* commands run in an 
interactive tty yield color output.

> Some sites, based on observation domain, would probably want to use L2 ethernet addresses,
> or source MPLS or VLAN tags to say what is local.  And I suspect some would like to have
> an extensive list of CIDR addresses, and want to put a " color to use " in the configuration.

Yes, extensive list of CIDR addresses + a " color to use " in the 
configuration would be great.

> I suspect that coloring the network part, and the host part of an address different colors would
> be interesting… but not sure if that is useful.

Nah, I don't see a scenario where this kind of coloring would be needed.

> So, if there are opinions on color, and how to use it, I'd love to hear about it before
> I release the color version of ratop().

Personally I've been using color like this:
* light green = IP belong to the customer and is site-local to the argus 
sensor (or in some times local to the person creating the report).

* green = IP belong to the customer but is remote, like a server at a 
remote office

* red = IP does not belong to the customer, i.e. it is external.

* purple = IP should be treated as if it were external, like the inside of 
a NAT:ing loadbalancer.

* cyan = excluded IP



Examples:
My customer has two offices and one DC.
10.10.0.0/16 = Office A
10.20.0.0/16 = Office B
172.16.0.0/24 = Private DMZ at DC
111.111.111.0/24 = Public DMZ at DC

A GET-request to the internet from a client at Office A:
The argus sensor is listening to traffic at Office A.
10.10.0.50 -> 123.123.123.123 80
light green       red


A GET-request to a server at DC, from a client at Office A:
The argus sensor is listening to traffic at Office A.
I haven't bothered with coloring public and private customer IPs 
differently.
10.10.0.50 -> 111.111.111.10 80   or
light green      green
10.10.0.50 -> 172.16.0.10 80
light green      green


A GET-request to a server at DC, from a client on the internet:
The argus sensor is listening to traffic at DC.
123.123.123.123 -> 111.111.111.10 80
      red               green


An incoming GET-request from the internet to a web-server on the private 
DMZ has its src IP source-NATed by the lb like this:
Original packet: 123.123.123.123 -> 111.111.111.10 80
After the lb:    172.16.0.1 -> 172.16.0.10 80

The argus sensor is listening to traffic at DC (inside the lb).
172.16.0.1 -> 172.16.0.10 80
   purple      light green
Even though this traffic physically come from 172.16 and go to 172.16, 
i.e. from and to the customer, external configuration stating that 
172.16.0.1 should be treated as if it was an external host colors this 
flow as purple -> light green instead of light green -> light green.


A customer network technician login to my customer portal. In his 
account profile it says he belongs to Office B. When he looks at the 
reports, IPs local to him (i.e. Office B) is shown in light green while 
Office A and DC IPs are colored in normal green.


Cyan should normally not be seen, but in some cases the customer SPAN is 
forced to include some garbage that you really don't want to see (and that 
you won't filter out via bpf). I simply color this in cyan to distinguish 
this traffic from the green and red sides.


So everything that has to do with the customer traffic use different 
shades of green and everything that has to do with the evil Internet use 
red shades throughout all reporting systems.



That's how I roll... :-)

/Elof


More information about the argus mailing list