Recently I was installing a kubernetes cluster as I usually do for my tests. However, as those machines were bare metal servers that some colleagues have recycled, we decided to keep it running and try to maintain it by ourselves.

First thing I did was a simple bot to send the alerts to a telegram channel. That was something I did not do in the past, because I do not care about monitoring as my clusters were ephemeral.

Last month I was picking my brain about GitOps and how this model fits with other kubernetes technologies like operators and backups. I decided to give it a try with ArgoCD. I created a private repo on GitHub, and started to set up everything. Suddenly, a thought came into my mind: I cannot store sensible information in a GitHub repo even if it is private and for testing purposes. So I started to dig into the alternatives using the links in the documentation.

Recently I have been asked a lot of similar questions like “does it make sense to invest in deploying legacy technology on kubernetes?”.

Well, first of all, I have to say that some technologies that are considered “legacy” have tons of new exciting features for the new workloads and applications. This is, for example, the case of DB2 with spatial queries and many others that you may have a look at (to be honest, I have very few experience with it).

Spring Boot is extensively used for microservice architectures, most of them running in kubernetes. Spring Cache has been defined as an abstraction layer for cache servers, although there are other alternatives (JSR-107 JCache was one of the initial ones).

When using a cache, there are two alternatives, embedded in the microservice or external to it. Both has its benefits and drawbacks and depending on the use case one is prefered to the other.

Infinispan is a java based open source cache and grid solution, that is used extensively in different framework like j2ee, springboot and quarkus among others. It is compatible with different java caching models, like JSR-107 and spring data.

It is very simple to set up a high availability cluster with different replication/distribution configurations, and thanks to the kubernetes operator pattern, is even easier in kubernetes distribution.

Prometheus is the defacto standard for metric scraping and short-term storing to identify potential issues. It is a graduated project of the CNCF Foundation.

I love Kubernetes and Ansible. Both are some of the 10 most popular Open Source projects in github in 2019. That is why I wanted to give it a try to deploy ansible awx in kubernetes, along with an ELK stack (Elasticsearch, Logstash and Kibana), to store playbook information and create some nice dashboards.

There are several ways to extract information from AWX. You can see some of them if you prefer the ELK stack in the following picture.