Hermes Wiki

ChaosEngineering

Deliberately injecting failure into a system to validate resilience assumptions actually hold.

Why we need this / what value this brings

Assumptions about resilience ('retries will handle it') are unverified until you've actually tested the failure.

When to use this

Once the system is mature enough that failures matter and there's a resilience story worth validating — far future for Localz currently.

How to use or implement this

Start by reasoning through failure scenarios by inspection (what happens if the DB connection drops mid-request) before ever running actual chaos experiments in production.

Research questions

  • Far future for Localz — but the underlying question ('what happens if the DB connection drops mid-request?') is worth answering by inspection now.

Empty folder — drop notes, links, and findings here as you research.

Hermes Wiki