You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@karaf.apache.org by Chris Geer <ch...@cxtsoftware.com> on 2012/11/21 17:41:54 UTC

Karaf Cellar - Offline containers and config updates

I've been looking at Karaf Cellar and have a question regarding
configuration changes and offline containers. My understanding is that
configuration changes are propagated amongst the containers through the
Hazelcast topics/queue. Is there anything in place to handle the case where
one container was offline/disconnected from the Hazelcast cluster during a
config change to re-sync once it connects again?

Thanks,
Chris

Re: Karaf Cellar - Offline containers and config updates

Posted by Jean-Baptiste Onofré <jb...@nanthrax.net>.
About queue yes and no ;)

For instance, a cluster:feature-install will do:

- update of the distributed map (containing the list of features per 
cluster group with the current status)
- broadcast the cluster event (using a Hazelcast topic)

However, in Cellar core, we provide a "wrapper" around queue that you 
can use. So we can leverage it (with some handler) for some action (we 
already use it for some administration commands).

Regards
JB

On 11/21/2012 05:59 PM, Chris Geer wrote:
> Thanks JB. I'll just have to take a look at the code at some point to
> fully understand it. I'll check it out after you do the release for 2.2.5.
>
> Is there any way to "queue" multiple changes so they all take place at
> once? That is probably closer to the Fuse Fabric profile concept though.
>
> Thanks,
> Chris
>
> On Wed, Nov 21, 2012 at 9:50 AM, Jean-Baptiste Onofré <jb@nanthrax.net
> <ma...@nanthrax.net>> wrote:
>
>     Hi Chris,
>
>     In Cellar, you have the Synchronizers that pull/push the
>     configuration (and the others resources).
>
>     You have two cases:
>
>     1/ when a new node join a cluster group, the synchronizer is called
>     2/ if a node leaves Cellar (for instance, network issue) and get
>     "outdated" regarding the distributed configuration and resources
>     states, you have to refresh cellar to call the synchronizer again
>
>     For 2/, I plan to add a cluster:sync command (and the corresponding
>     MBean) to "force" the call to the synchronizer.
>     It will be included in Cellar 2.2.5 (that I plan to release tonight ;)).
>
>     Regards
>     JB
>
>
>     On 11/21/2012 05:41 PM, Chris Geer wrote:
>
>         I've been looking at Karaf Cellar and have a question regarding
>         configuration changes and offline containers. My understanding
>         is that
>         configuration changes are propagated amongst the containers
>         through the
>         Hazelcast topics/queue. Is there anything in place to handle the
>         case
>         where one container was offline/disconnected from the Hazelcast
>         cluster
>         during a config change to re-sync once it connects again?
>
>         Thanks,
>         Chris
>
>
>     --
>     Jean-Baptiste Onofré
>     jbonofre@apache.org <ma...@apache.org>
>     http://blog.nanthrax.net
>     Talend - http://www.talend.com
>
>

-- 
Jean-Baptiste Onofré
jbonofre@apache.org
http://blog.nanthrax.net
Talend - http://www.talend.com

Re: Karaf Cellar - Offline containers and config updates

Posted by Chris Geer <ch...@cxtsoftware.com>.
Thanks JB. I'll just have to take a look at the code at some point to fully
understand it. I'll check it out after you do the release for 2.2.5.

Is there any way to "queue" multiple changes so they all take place at
once? That is probably closer to the Fuse Fabric profile concept though.

Thanks,
Chris

On Wed, Nov 21, 2012 at 9:50 AM, Jean-Baptiste Onofré <jb...@nanthrax.net>wrote:

> Hi Chris,
>
> In Cellar, you have the Synchronizers that pull/push the configuration
> (and the others resources).
>
> You have two cases:
>
> 1/ when a new node join a cluster group, the synchronizer is called
> 2/ if a node leaves Cellar (for instance, network issue) and get
> "outdated" regarding the distributed configuration and resources states,
> you have to refresh cellar to call the synchronizer again
>
> For 2/, I plan to add a cluster:sync command (and the corresponding MBean)
> to "force" the call to the synchronizer.
> It will be included in Cellar 2.2.5 (that I plan to release tonight ;)).
>
> Regards
> JB
>
>
> On 11/21/2012 05:41 PM, Chris Geer wrote:
>
>> I've been looking at Karaf Cellar and have a question regarding
>> configuration changes and offline containers. My understanding is that
>> configuration changes are propagated amongst the containers through the
>> Hazelcast topics/queue. Is there anything in place to handle the case
>> where one container was offline/disconnected from the Hazelcast cluster
>> during a config change to re-sync once it connects again?
>>
>> Thanks,
>> Chris
>>
>
> --
> Jean-Baptiste Onofré
> jbonofre@apache.org
> http://blog.nanthrax.net
> Talend - http://www.talend.com
>

Re: Karaf Cellar - Offline containers and config updates

Posted by Jean-Baptiste Onofré <jb...@nanthrax.net>.
Hi Chris,

In Cellar, you have the Synchronizers that pull/push the configuration 
(and the others resources).

You have two cases:

1/ when a new node join a cluster group, the synchronizer is called
2/ if a node leaves Cellar (for instance, network issue) and get 
"outdated" regarding the distributed configuration and resources states, 
you have to refresh cellar to call the synchronizer again

For 2/, I plan to add a cluster:sync command (and the corresponding 
MBean) to "force" the call to the synchronizer.
It will be included in Cellar 2.2.5 (that I plan to release tonight ;)).

Regards
JB

On 11/21/2012 05:41 PM, Chris Geer wrote:
> I've been looking at Karaf Cellar and have a question regarding
> configuration changes and offline containers. My understanding is that
> configuration changes are propagated amongst the containers through the
> Hazelcast topics/queue. Is there anything in place to handle the case
> where one container was offline/disconnected from the Hazelcast cluster
> during a config change to re-sync once it connects again?
>
> Thanks,
> Chris

-- 
Jean-Baptiste Onofré
jbonofre@apache.org
http://blog.nanthrax.net
Talend - http://www.talend.com