How to handle availability zones and failover with ansible? Is it even within the scope of ansible?

Question

I am only getting started with ansible.

I want to set up a new development infrastructure with the infrastructure as code principle. First thing I'll have to prove is a Jenkins instance with a few nodes for a Java dev team.

I want to do it correct right from the start, so I want to include some redundancy. Let's assume that in this scenario there are 3 sites with bare metal servers. I want to set up a Jenkins master in all 3 zones which form a cluster.

But I don't want to think about that myself. I just want to provide the info to ansbile what server is in which availability zone, and ansible should make sure to set up a Jenkins master in each availability zone and connecting it to the cluster.

Maybe this is already out-of-scope for ansible and I need another tool on top of ansible?

Any thoughts?

Answer 1

What your are describing can typically be addressed with a correct inventory design.

[zone1]
serverA
serverB
serverC

[zone2]
server1
server2
server3

[zone3]
serverX
serverY
serverZ

[jenkins_master]
serverA
server1
serverX

[jenkins_slave]
serverB
serverC
server2
server3
serverY
serverZ

From there you can easily:

• create plays that will target a group or a pattern and that you can limit to an other group or pattern at runtime, e.g. target jenkins_master and run the playbook on all of them or limit it to zone1.
• create inventory vars for your hosts and groups so that e.g. servers in zone1 will know what is the uri of the master in that zone.

This is a super-quick overlook. I suggest you read in depth working with inventory to get a better understanding of all the inventory concepts and pay attention to the part concerning the organization of variables in the inventory for groups and hosts