Re: [cf-dev] Do you use CoapClient API?

["Kovatsch Matthias" <kovatsch@xxxxxxxxxxx>]

Hi there

If you need the merge to make progress, we can do it. I simply want to stress that we will not be done and probably touch this part of the code again. The distributed Matcher/ObservationStore is fine. However, I see quite some problems at the upper end, where you currently have this list of NotificationListeners. Personally, I don't think the architecture is intuitive. Thus, we will discover a lot of issues with getting the consistency right (notification ordering and correct application behavior).

Kay, will Simon's PR also suit your needs, since you have this alternative proposal? If yes, please merge.

Simon, be ensured that I am fully on board to get nice horizontal scalability! Also, I am super thankful for the work you guys put into the project. Yet please keep in mind, that we will have "work in progress" code after merge. The tough issues are still not addressed and I think it will become easier without this notification bypass. (I think the notification orderer must be similar to the ObservationStore. It must sync between all receiving handlers to ensure consistency. Since this is a point where intra-node communication is needed, it would also be a good place to solve the fail-over issue and dispatch to a live handler on one of the nodes.)

Ciao
Matthias

________________________________________
From: cf-dev-bounces@xxxxxxxxxxx [cf-dev-bounces@xxxxxxxxxxx] on behalf of Kai Hudalla [sophokles.kh@xxxxxxxxx]
Sent: Thursday, June 02, 2016 11:06 AM
To: Californium (Cf) developer discussions
Subject: Re: [cf-dev] Do you use CoapClient API?

>>
>> Simon has tried to find a way to support fail over of observations for
>> both types of API, thus he registers a listener for each instance of
>> CoAPClient to be notified when notifications come in that have been
>> created on a different node. Personally, I don't think that we need this
>> and that we instead should provide the fail over behavior for the
>> Endpoint API only.
> I don't really try to support the fail over for CoapClient.
> I changed the endpoint API to support failover and clustering.
> This impact the coapClient implementation so I changed it to keep the
> same behavior as before.
> Nothing more.

Thanks for the clarification, Simon. I didn't want to put words into
your mouth :-) See you next week.
_______________________________________________
cf-dev mailing list
cf-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cf-dev