Hi All,
Im looking for some implementation ideas where i can use information from the system.compute.node_timeline table to catch memory spikes and if above a given threshold restart the cluster through an API call.
Have any of you implemented a similar solution? Any references would help.
