Getting your first application ready for production and preparing a production cluster for its live debut is an important step in the container-adoption process. This is where we see rookie mistakes damage the reputation of very smart technologists. In one case, without running High Availability (HA) for non-production, the customer built their first HA cluster for production, not practicing with the non-production cluster first and did it with just two managers nodes. Had they started in non-production the failure would have been much less visible (remember, an even numbers of managers means trouble)! Some key lessons learned here were to keep your non-production and production cluster architectures very similar and always try new things in the non-production cluster first.
In this chapter, we will look at important topics...