[ejabberd] node_type and frontend/backend separation.

Tim Stewart tim at stoo.org
Mon Mar 9 18:01:36 MSK 2009


On Mar 9, 2009, at 4:31 AM, Mickael Remond wrote:

>> I came across the option and
>> the logic in ejabberd_node_groups.erl while looking for a way to
>> restart an individual ejabberd backend without interrupting IM client
>> connections.  Is this what this option is designed to do?
> There is no logic for that in the main code, but it could be added.
> The logic behind this architecture is described there:
> http://www.process-one.net/en/blogs/article/angie_introducing_flexarch

I see that you're referring to this:

"It is actually possible to write some code that keep the XMPP TCP  
connexion on the front-end but could survive to a loss of the back end  
managing its session (by “migrating” the process)."

I'm on a deadline and I'm not sure how involved implementing this  
process migration would be.  Is it just be a matter of detecting when  
one backend has failed or been shut down, finding a new backend, and  
retrying the operation?  Does ejabberd already use the code in  
ejabberd_node_groups.erl, and it just hasn't been extended to support  
this flexibility yet?

Thanks again for the help,


Tim Stewart
tim at stoo.org
+1 404 475 4848

More information about the ejabberd mailing list