Question About Consumer Status Checks

[palm002]

Summary

Hi there!

I have a question regarding the consumer status checks (isRunning and isPolling) (reference). We're looking to base our
Kubernetes liveness probe on the consumer status, but I've found the documentation a bit unclear about whether we
should be using the isRunning or isPolling status. Could you advise?

For context, so far we've tried using isPolling, however have observed cases where isRunning is true while
isPolling is false. What would the reason for this be? Could you help me better understanding the differences
between "running" and "polling."

Thanks!

PS: I am using node v18.20.4 and "sqs-consumer": "^11.2.0" if that is relevant

Example

No response

[nicholasgriffintn]

So polling is only true on the schedule that you have set.

Running is if start has been called and stop hasn't.

Does that make sense?

[palm002]

Thanks for your response! I appreciate your clarification on isRunning and isPolling. However, I'm still a bit unsure about how to determine if the consumer is healthy and connected to the queue. We don't use consumer.stop() explicitly but do have a process that kills kubernetes pods in order to refresh credentials. This is expected behaviour from our side. Furthermore, we don't provide a polling schedule explicitly.

Could you please explain what indicators we should look for to confirm the consumer's health? Based on that, which status—isRunning or isPolling—would be best for our Kubernetes health checks?
Cheers!

[nicholasgriffintn]

Hey, sorry for the delayed response, I had a lot of notifications.

So is running is the check that will tell you if the consumer is running and hasn't stopped, that can be used as an indication of health, it's not an indication of full health though to note, only that the consumer is running.

Is polling is only an indication of when the consumer is polling, this happens on a schedule based on the configuration provided and the defaults.

[palm002]

No worries. Again, thanks for you reply and providing further clarifications.
It sounds like there's no black and white answer here. I might settle on isRunning check for now given that I see lots of errors on my k8s liveness probes when utilizing isPolling since that only returns true based on the schedule.
Happy to close this thread and mark as resolved since it clarifies my initial confusion related to documentation.