Idea #13921
Updated by Tom Morris over 5 years ago
The original ticket expressed the need in terms of container requests, but I think we've got that piece reasonable well covered with our existing logging, so I'd like to repurpose this to do logging of stats from the point of view of the Node, not the Container. When a node is shut down, record how long it was up as well as times for: booting, downloading docker images, running docker images, and idle. This should allow system administrators to monitor overall performance, efficiency, and cost of the system orthogonally to jobs being run (which we account for through the container logs). -For For billing purposes, we want to know the time between when the node is up and ready to be used and when crunch-run is running, as well as when crunch-run has stopped and when the node is no longer available (or has moved on to another container_request)- container_request)