In the summer we wrote about how CloudNativePG was back for the September- October-November LFX term with several projects for mentoring. One of them was around Chaos Testing.

Yash Agarwal worked with mentors and CloudNativePG maintainers Gabriele Bartolini, Marco Nenciarini, Francesco Canovai, and Jonathan Gonzalez, to enhance the project’s test coverage. Introducing LitmusChaos, a comprehensive chaos testing framework, the team designed automated chaos experiments for common failure scenarios, integrated them into CI/CD workflows, and collected observability metrics like failover time and data consistency. I had a chat with Yash about his work, and about how he got into Tech in the first place.

Start at the beginning

Yash’s venture into programming started when he got introduced to Python in 11th grade. He was always fascinated by technology, and got further inspired to pursue a career as a programmer by his cousin Amit, a software developer.

Today Yash is a full stack developer intern at Seeqlo, where he, among other things, focuses on streamlining cloud operations and optimizing performance. Based in Bengaluru, India, Yash is a member of Point Blank, a student-run tech community dedicated to learning together.

He looks back at working with the CloudNativePG team as a “great learning experience”. They met twice a week for 30 minutes to discuss the progress of the project. One thing that Yash says he learned is to have more patience.

Chaos testing

The new chaos-testing repository Yash worked on provides automated tools to validate PostgreSQL cluster resilience under failure conditions. It combines two testing approaches:

• Jepsen Consistency Testing - Uses the famous Jepsen framework to perform mathematical proofs of database consistency. It continuously runs database operations (50 ops/sec) and validates that no data is lost or corrupted during failures.
• LitmusChaos Fault Injection - Uses LitmusChaos to simulate real-world failures by repeatedly deleting the PostgreSQL primary pod (every 60-180 seconds), forcing CloudNativePG to perform automatic failover.

You can read more about the project in the repository’s README. And, in case you’re curious, here’s Yash’s PR: https://github.com/cloudnative-pg/chaos-testing/pull/3

Contributing to Litmus itself

Yash wasn’t able to find how to get the chaos engine to target the primary pods since the appKind CloudNativePG uses isn’t natively supported by Litmus. “I tried many things, but when I tried AppKind as “cluster” it worked! I read the Litmus code and found that there were some validations which prevented “Cluster” (capital “C”) from working. This behavior was not described in Litmus’ documentation, which meant I could submit a PR and prevent the next person from running into the same issue!”

What’s next?

In the second half of his 3rd year, Yash is exploring opportunities in the field of backend and DevOps. “I will surely try to contribute more towards CloudNativePG when time permits!” You can follow Yash’s work on GitHub.