Development discussion of WireGuard
 help / color / mirror / Atom feed
From: Damian Kaczkowski <damian.kaczkowski@gmail.com>
To: "Jason A. Donenfeld" <Jason@zx2c4.com>,
	WireGuard mailing list <wireguard@lists.zx2c4.com>
Subject: Re: Ability to use one udp port for multiple wg interfaces
Date: Fri, 5 May 2017 20:28:33 +0200	[thread overview]
Message-ID: <CAGzienGzypKvYQZBShJuZp9kZHhj26Rc8-qtObD+Ap_+XhB4+w@mail.gmail.com> (raw)
In-Reply-To: <CAHmME9rReXuDHmGTU8k7+ryu3pWSMHR0aGfaFWQzWhAnZy9DjA@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 2652 bytes --]

On 2 May 2017 at 21:45, Jason A. Donenfeld <Jason@zx2c4.com> wrote:

> On May 2, 2017 19:59, "Damian Kaczkowski" <damian.kaczkowski@gmail.com>
> wrote:
>
> On 2 May 2017 at 18:32, Jason A. Donenfeld <Jason@zx2c4.com> wrote:
>
>>  > 3. Well if one uses firewall to control flows between zones in
>> environment
>
> > with mix protocols (eg. gre, ipsec, openvpn and so on) then using second
>> > tool just to control only wireguard ACLs is not very convenient way from
>> > administrative point of view. Also in case where peer is roaming and
>> > changing its source IP (eg. road warrior) then maintaining wireguard
>> ACLs
>> > will be a huge PITA, if not impossible at large scale.
>>
>> No, you are wrong. Allowed-ips controls the IP addresses _within_ the
>> tunnel. Thus your iptables rules can use "-i wg0 -s 10.0.0.3/32" or
>> similar to match a _precise_ peer.
>>
>
> Ok. Thanks for a tip. However I still think wireguard looses some
> flexibility in that way eg. when peer roams from one network to another
> then its ip address may be unknown.
>
>
> No, wrong. Roaming regards external IP. Allowed IPs regards internal
> tunnel IPs, which are static.
>

True. But I still think that ability to assign multi interfaces to one udp
port would be handy. Eg when one want to use only specific and limited
ports (like eg 53) for wireguard but still wants to have more interfaces at
one's disposal. Possible use case - be able to easily assign and group
various peers to different interfaces and monitor those interfaces
parameters (exposed eg by kernel) using monitoring tools capable of
collecting various information/data and/or plotting graphs like bandwidth,
traffic, and so on. This info could be later used for analysing or
debugging.


Anyway, it is not only about roaming case so if it is not much of a work
> and if it is not a security problem then please consider to allow multiple
> wg interfaces to work on one port. I hope it won't hurt to allow this
> functionality and I am sure it might come handy for some admins in the
> wild. Maybe it could be implemented in pair with the idea of refactoring
> per interface vs per peer private keys? Hope you will consider it at some
> point.
>
>
> No, you are very mistaken. Please reread the docs on allowed ips keeping
> in mind that these concern internal tunneled ips and are static. Typing to
> you on my phone so can't write more now.
>

Reading through the docs.

I come to scenario where I would like to disable whole allowed-ips thing. A
multi-homed scenario where traffics comes out via one peer/interface and
come back via the other one. Wrote about this in another email.

Greets.

[-- Attachment #2: Type: text/html, Size: 4833 bytes --]

  reply	other threads:[~2017-05-05 18:19 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-05-02  8:20 Damian Kaczkowski
2017-05-02  8:55 ` Jason A. Donenfeld
2017-05-02  9:56   ` Damian Kaczkowski
2017-05-02 16:32     ` Jason A. Donenfeld
2017-05-02 17:59       ` Damian Kaczkowski
2017-05-02 19:45         ` Jason A. Donenfeld
2017-05-05 18:28           ` Damian Kaczkowski [this message]
2017-05-11 10:30             ` Jason A. Donenfeld

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=CAGzienGzypKvYQZBShJuZp9kZHhj26Rc8-qtObD+Ap_+XhB4+w@mail.gmail.com \
    --to=damian.kaczkowski@gmail.com \
    --cc=Jason@zx2c4.com \
    --cc=wireguard@lists.zx2c4.com \
    /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;
as well as URLs for NNTP newsgroup(s).