"AggregatingTest Results from Multiple Pods with k6-operator-system: Any Recommended Approaches?

We are currently using the k6-operator-system with parallelism: 2 for our tests. However, we’re encountering difficulties in aggregating the test results from all pods accurately. While we understand that aggregation can be done at Prometheus, it’s not entirely accurate for percentile metrics like latency. Our request is to aggregate metrics at the source, i.e., within the k6-operator itself.

Any insights or suggestions would be greatly appreciated.

Thank you!

Hello @saik6,

Welcome back to the community forums! And apologies for the delayed answer :pray:

Personally, I don’t have any experience with the k6 operator, at all.

But, perhaps its main contributor and our k6 operator expert @olhayevtushenko can help, as I guess they have way longer experience using it.

Apologies! :bowing_man:

Hi @saik6 and @joanlopez!

k6-operator does not aggregate metrics, and I’d say it is very unlikely it is ever going to :sweat_smile: Metrics aggregation is a rather non-trivial task and as you mentioned, there are tools that are specialized in that area like Prometheus. Or our cloud solution for that matter.

While we understand that aggregation can be done at Prometheus, it’s not entirely accurate for percentile metrics like latency.

Could you share details on that? It might be that this issue came up before, here on the forum or in Github repos (e.g. in Prometheus extension repo or in k6 repo). If so, additional feedback might help.

IOW, I’d suggest to try to dig into this instead.

Hope that helps!

1 Like