On Thursday, April 30, 2015 at 12:10:51 PM UTC-5, Jochen Haeberle wrote:
>
> Hi Cesar,
>
> I know of no way to synchronise catalog compilation.
>
>

Me neither.  However, you can control which resources are managed for each 
node, and which properties are assigned to them.

 

> I would try to update all nodes in a first step. When all agents are 
> updated, I would change the IP of the puppetmaster.
>


Sounds like a plan to me.  Specifically, the nodes get updated with both 
their own new IP addresses and the new (not yet implemented) IP for the 
master.  Address lookups for the master (and for other nodes not yet 
updated) will return the wrong result for a time, but I can't think of any 
strategy with or without Puppet that could altogether avoid such a result.  
Once all the agents have updated (or at least retrieved their catalogs 
successfully), the master updates itself.

You could orchestrate the master vs other agents timing manually, but if 
you are willing to rely on all agents to update successfully within a 
predictable window, then you could consider using Schedules to 
automatically control when the master's update can take place.

 

> If the nodes do not reconnect to the master automatically, because for 
> example the agent stops working, I would add a cron job to the agents to 
> keep them trying. This would require a third step to remove the cron job 
> afterwards…
>
>

Or if you run the agent out of cron anyway then you get that for free.  
Determining what you might need to do in this area certainly requires 
testing; don't skip that!


John

-- 
You received this message because you are subscribed to the Google Groups 
"Puppet Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/puppet-users/baa91f24-bffa-47f2-ae39-b6ad68104c7c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to