Hi Robert,

> On 15 Jan 2018, at 12:23, Robert Kisteleki <[email protected]> wrote:
> 
> On 2018-01-15 13:09, Tim Chown wrote:
>> Hi,
>> 
>> At 
>> https://atlas.ripe.net/about/faq/#so-which-services-do-i-need-for-my-probe-to-work
>> 
>> it says
>> 
>> "The absolute minimum set is DHCP, DNS and outgoing TCP port 443 (HTTPS) in 
>> order to allow the probe to connect to the network. However, this in itself 
>> is not enough to do measurements, which is the entire focus of RIPE Atlas, 
>> so you should also allow ICMP, UDP (DNS + traceroute), TCP for traceroute 
>> and HTTP(S)."
>> 
>> What specific ports and protocols are required for routine operation and for 
>> inbound or outbound measurements to be configured?  I think the above info 
>> could be a little more detailed (having had questions asked of me).
>> 
>> Many thanks,
>> Tim
> 
> Hi,
> 
> The more precise we try to be, the more wrong we'll end up being :-) but
> I'll try to be a bit more specific.
> 
> For incoming traffic: the probes don't provide real accessible services,
> so incoming ICMP/ping and UDP/traceroute is probably enough (assuming
> the probe is otherwise not firewalled / NATed).

I think some probes we'd like to run tests to are behind a firewall, hence the 
interest on what's required as a minimum for at least basic connectivity tests. 
 I'll follow up directly with a couple of specific examples rather than cite 
them here.

> For outgoing traffic: the more you allow, the more measurements will
> have a chance of succeeding. For example, if you only allow TCP/443 out,
> then measurements to other ports (like TCP/traceroute or TLS to non-443)
> will likely fail. Allowing outgoing DNS to any server is a must in order
> to be useful for non-local-resolver queries. And so on.

OK, thanks.  A tweak to the FAQ along those lines would be good, I think :)

> We also have NTP since the writing of the above FAQ entry, and HTTP
> towards anchors. While the requirements (or, I should say,
> recommendations) don't change each day, they do evolve over time.

Understood, and thanks again.

Tim

> 
> Hope this helps!
> Robert
> 


Reply via email to