Testing configuration and deployment with some config management tool


I am planning to configure HAProxy through Puppet in order to keep configuration consistent between nodes, but my concerns are, how can I verify the syntax that I set in Puppet is right before deploying to production, and also how would I handle the service reload without knowing when the puppet recipe is going to be applied?

So, how would the workflow be ?



You can check the syntax with the “-c” argument.

I don’t see how we would be able to respond regarding your question about how the puppet workflow needs to look like. That is something you need to know figure out based on your own requirements.


Right, but I wanted to say how the haproxy config update would be, for instance:

I have 2 nodes (active-passive )
I apply the new config on the passive
Reload the passive
Apply the same config to the active
Reload it (would it failover the passive?) or maybe it is better to just failover the backup one and then apply the config there?

This is what I mean about the workflow. Also when managing those configurations with a tool like puppet, which has a daemon that will run at an unknown time…

So I am just wondering how to handle those cases. Should I stop puppet before applying any config and then pull changes from puppet repo one by one and in a control manner ?
Don’t know if someone else has face this situation.



Update whenever you want, whatever instance you want and reload it.

Reloading is less intrusive than failing over to the standby instance. No reason to over complicate things.