Conference kernel::csguk_systems

76.0. "Diagnosing Hung Systems" by KERNEL::JAMES (Alan James CSC Basingstoke) Wed Oct 18 1989 14:38

		In the good old days diagnosing hung systems was a doddle.
	Today we have CI's and clusters, and the associated timers.

		What is the best way to diagnose a hung system in a cluster?

		Much depends on the circumstances. What most Engineers are
	doing is :-

			Halting processor   |
			Examining PSL       |      very fast !!!
			Continuing          |

                           then repeating this process many times to obtain
	                       a general idea of where the PC and PSL lie.

	After this they sometimes attempt other commands before forcing a crash.

                If you have discovered useful techneques in gathering
	information from hung systems then perhaps you would be good enough
	to add them to this notes entry.

	Alan.