I think it's not as useful to track this as an average over time or at least I don't see which questions this helps to answer.
What we might be interested in is the event loop utilization (ELU) during a certain time frame of the slot, e.g. 1-4, 4-8, 8-12, or even on a per second basis. Should be able to achieve this via labels on metrics + running an interval on a clock timer.
This could help us analyze what parts of the slot we are overloaded, and if a change improves the ELU for a specific slot frame or not.
I think it's not as useful to track this as an average over time or at least I don't see which questions this helps to answer.
What we might be interested in is the event loop utilization (ELU) during a certain time frame of the slot, e.g. 1-4, 4-8, 8-12, or even on a per second basis. Should be able to achieve this via labels on metrics + running an interval on a clock timer.
This could help us analyze what parts of the slot we are overloaded, and if a change improves the ELU for a specific slot frame or not.
Originally posted by @nflaig in https://github.com/ChainSafe/lodestar/issues/6820#issuecomment-2190189746