[zeromq-dev] Reundance/Failover Capabilites?

[Martin Sustrik]

Hi Evan,

> My question is about the semantics 0MQ provides for redundancy/failover. From
> what I can tell, there are currently none -- is that correct? I couldn't see
> anything regarding this in the documentation, but there's also a bullet point
> on the front page stating that a goal is to be decentralized, which implies
> that this might be in the works?

I assume you are speaking about applications reconnecting after 
network/application failure rather than about guaranteed delivery.

Currently, when peer disconnects, you are notified about the event via 
callback function and you can reestablish the connection if needed.

This is rather unconvenient, therefore we are working on automating it 
(should be available in version 0.4). The idea is that applications 
won't be even notified about connection failures and 0MQ will reconnect 
automatically as soon as possible. The messages sent by the application 
will be queued in the meantime and actually sent to the network 
immediately after reconnection.

Martin