[zeromq-dev] ZBeacon multiple interface support
Arnaud Loonstra
arnaud at sphaero.org
Fri Apr 10 10:21:03 CEST 2015
I'm not sure if I get what you mean but I think there exist many
networking solutions which support multiple interfaces. From the
perspective of Zyre I was thinking of registering multiple endpoint
addresses of nodes and then use these based on priority. If one endpoint
goes down the second is tried until all registered addresses are tried
to finally purge the node.
I guess the zyre_peer class's endpoint member would then become a list
of endpoints as well as the mailbox member. Question remains how
priorities would be determined. Could be done by interface metrics?
Rg,
Arnaud
On 2015-04-10 08:37, Pieter Hintjens wrote:
> Once you have multiple interfaces, you need to think about bridging
> traffic between them, if your intention is to create a single ZRE
> network. It goes beyond beaconing therefore. I think you need one
> instance of zbeacon for each interface, and then explicit code to
> bridge the two TCP networks. I assume it gets complex when you have
> multiple devices acting as bridges.
>
> Since it's a problem we never faced yet, nor had to solve, it remains
> a theory.
>
> On Thu, Apr 9, 2015 at 4:03 PM, Arnaud Loonstra <arnaud at sphaero.org>
> wrote:
>> Hi All,
>>
>> I was refactoring the ZBeacon code in Pyre to start thinking about
>> using it on multiple interfaces. Imagine a simple laptop with a
>> wireless
>> and wired NIC.
>>
>> ZBeacon now tries the first interface it can get its hands on or it
>> tries INADDR_ANY.
>>
>> To support multiple interfaces it would need to transmit the
>> ipaddress
>> of the interface it is broadcasting on. Then Zyre would also need to
>> support multiple endpoint addresses of nodes.
>>
>> I can imagine having the zyre_node class start zbeacon instances on
>> every interface it finds. Or have zbeacon itself open sockets on
>> every
>> interface. I don't know what would be most suitable approach?
>> Especially
>> with networks being plugged in and out...
>>
>> Has anybody already thought this true? Any suggestions?
>>
>> Rg,
>>
>> Arnaud
>> _______________________________________________
>> zeromq-dev mailing list
>> zeromq-dev at lists.zeromq.org
>> http://lists.zeromq.org/mailman/listinfo/zeromq-dev
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> http://lists.zeromq.org/mailman/listinfo/zeromq-dev
More information about the zeromq-dev
mailing list