We deploy our code on kubernetes and have integrated the circuit breaker for redis in our ruby on rails using semian gem. In the document, it says that the circuit breaker feature is not coordinated server wide.
- Does it mean that if the circuit opens in one pod in one of the deployment, then it may not break in other pods in the same/another deployment?
- If yes, then does it mean that I need to calculate the number of requests made by each pod? So, let's say the total redis calls made per minute by the system is 1000 and there are 10 pods in the system. Hence, ideally 1 pod makes 1000/10 = 100 req/minute. So, to break a circuit, I have kept the error threshold value as 50 and error timeout value as 60 seconds. It means that if one pod makes 50 error calls in 60 sec, then the circuit will open only in that particular 1 pod and not in the other 9 pods unless other 9 pods also error out for 50 req in 60 seconds. Is my logic correct or am I missing anything here?
- Can I use the circuit breaker feature at a deployment level than the pod level?
