Investigate a Slurm node in down, drain or drng state on a GPU cluster. Use when a node disappears from the available pool, when capacity drops unexpectedly, or when asked why a specific node is not accepting jobs. Reads the state and the reason before assuming hardware.
