public inbox for development@lists.ipfire.org
 help / color / mirror / Atom feed
From: Michael Tremer <michael.tremer@ipfire.org>
To: development@lists.ipfire.org
Subject: Re: [PATCH] Add GeoIP location to nameservers
Date: Tue, 19 Jan 2016 01:30:39 +0000	[thread overview]
Message-ID: <1453167039.5665.159.camel@ipfire.org> (raw)
In-Reply-To: <56936382.1070008@web.de>

[-- Attachment #1: Type: text/plain, Size: 5611 bytes --]

Hi,

On Mon, 2016-01-11 at 09:10 +0100, IT Superhack wrote:
> Hello Michael,
> 
> Michael Tremer:
> > Hi,
> > 
> > On Sun, 2016-01-10 at 17:25 +0100, IT Superhack wrote:
> > > Hello Michael, hello Matthias,
> > > 
> > > Michael Tremer:
> > > > Just out of curiosity, why do you find this information so
> > > > helpful?
> > > 
> > > As Matthias said already, is is more a "nice to have" than
> > > something
> > > which is seriously needed.
> > 
> > This is not too much of an argument. My argument against this is
> > that
> > it brings down page load times because of a not too useful
> > information.
> > 
> > > I wrote this patch because a friend of mine in France discovered
> > > that
> > > his ISP assigns DNS servers from Australia and Great Britain,
> > > which
> > > was slowing down DNS resolving a lot.
> > 
> > I get that and this is actually a pretty good one.
> > 
> > That only leaves resolvers like 8.8.8.8 which will show "US" but
> > actually are located at many places around the world. Let's hope
> > that
> > people don't get the wrong thing from the flag - or actually start
> > changing their DNS servers to something else :)
> > 
> > > Therefore I thougt it might be useful to see in which countries
> > > your
> > > DNS servers are located, just in case you didn't set some by your
> > > own.
> > 
> > It is sometimes. Although geographic location doesn't mean that it
> > is
> > close on the network.
> > 
> > A system in GB is probably not an issue. Australia actually is a
> > bit
> > far away.
> The problem here was something else: A couple of months ago, his ISP
> assigned DNS servers in France, which worked quite well and belonged
> to the ISP, according to whois information.
> 
> The list of assigned DNS servers must have changed somewhere in the
> meantime, and he still does not know why since the ISP does not
> answer
> related questions.
> 
> So, if your ISP suddenly assigns you DNS servers in a very different
> location
> than it has done long before, you know that something might be wrong
> here.
> (The Quantumhand-program of the NSA lists "DNS injection" as a
> possible
> method to impersonate a server - why not even change the DNS server
> to
> one they own? Would be much easier...) 
> [http://securityaffairs.co/wordpress/23129/hacking/quantumhand-nsa-im
> personates-facebook-inject-malware.html]

Yeah, I get where you are coming from. The flag however won't change
much here. A check if the IP address of the DNS servers are part of the
AS of the ISP would be a better one to find fakes.

A latency check would find name servers that are "too far" away on the
network.

The GeoIP information just very loosely corresponds to a few of these
things. Nothing more in my opinion.

> > 
> > > In general, adding geographic information to IP addresses is very
> > > helpful in my point of view because anomalies can be detected
> > > much
> > > better and more precise firewall rules are possible.
> > 
> > I don't get why.
> For example, many active connections from an internal host to china,
> korea or an african country might indicate that a host is infected.
> If someone need to call ipinfo.cgi for every IP he/she/it does not
> know,
> it will end in a nightmare...

Nobody will check this at the right time. Something like snort would
help you more here.

> > 
> > > However, some thing might still be improved: For example, the
> > > ipinfo.cgi
> > > file shows the IP address, the rDNS name, whois information, but
> > > not
> > > the appropriate flag. So, if someone scrolls through the
> > > connection
> > > tracking
> > > page, he/she/it sees the source and destination IPs of any active
> > > (and
> > > recently closed) connection. At the moment, there is no way of
> > > telling
> > > which country an IP belongs to - without using additional web
> > > services, of
> > > course - since the flag is shown neither at the connection
> > > tracking
> > > page
> > > nor at the ipinfo.cgi page. This isn't very helpful, is it?
> > 
> > The ipinfo.cgi page shows the whois information for an IP address.
> > That
> > may contain the name and HQ location of a company this IP address
> > belongs to, but that does *not* mean that the host is actually
> > located
> > in that country - and almost certainly not at that address.
> That's right, GeoIP shows the location of the server and not those of
> its owner.

No, still wrong. Just check 8.8.8.8. GeoIP will show you US. It is
actually an anycast with many hosts in the world. That is quite typical
for name servers.

This *must* be added to the documentation in the wiki and state very
clearly that this information is never very accurate and especially
with name servers less accurate than usually to expect.

> But looking at this whois output:
> 
> CariNet, Inc. NET-26 (NET-71-6-158-128-1) 71.6.158.128 - 71.6.158.191
> CariNet, Inc. CARINET-5 (NET-71-6-128-0-1) 71.6.128.0 - 71.6.255.255
> 
> Not very helpful in first place, is it? ;-)

Why not?

> > 
> > The GeoIP database is a completely different thing.
> > 
> > Judging by the location of the host does make any sense if you care
> > about security.
> Basically, yes. As I mentioned above, GeoIP makes more sense to
> detect
> anomalies and to allow, e.g., only VPN access from countries which
> are
> necessary.

You should know that IPFire downloads this GeoIP database from an
untrusted source over an untrusted connection.

> > 
> > > 
> > > That is basically the motivation behind the two patches I
> > > submitted
> > > recently.
> > > 
> > > Best regards,
> > > Timmothy Wilson
> Best regards,
> Timmothy Wilson
> 

-Michael

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

  reply	other threads:[~2016-01-19  1:30 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <5686D08D.7000802@ipfire.org>
2016-01-04 16:59 ` Michael Tremer
2016-01-04 17:16   ` Matthias Fischer
2016-01-10 16:25   ` IT Superhack
2016-01-10 22:15     ` Michael Tremer
2016-01-11  8:10       ` IT Superhack
2016-01-19  1:30         ` Michael Tremer [this message]
2016-01-19  6:57           ` IT Superhack
2016-01-01 17:11 IT Superhack

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1453167039.5665.159.camel@ipfire.org \
    --to=michael.tremer@ipfire.org \
    --cc=development@lists.ipfire.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox