I am replying to my own message so if people search the archive they can figure this out faster than I did.
This was occurring because the event went green --> red --> yellow --> red And I did not have yellow set as a "okcolors" so the logic for A_OK in hobbitd was never called (clear the ack message) - but the logic for state change was (change cookie)
Dunno why you would want it to not clear the ack message but generate a new cookie, but adding yellow to the okcolors fixed it.
Thanks everyone for the suggestions
-Sean
-----Original Message----- From: Sean R. Clark [mailto:sclark at nyroc.rr.com] Sent: Monday, June 04, 2007 9:43 AM To: 'hobbit at hswn.dk' Subject: ACK problems
Odd ack problem
I have a machine that went red for CPU
It generated an "cookie" of 371004 , and paged reciepent1. It was then ACK'd for 8 hours. CPU recovered after 2 hours, then went red again. This generated a new "cookie" of 344975, which we could no longer ACK because in the hobbitdboard it had the "old" ack data and time. This new "cookie" then escalated (because we could not ack this event)
Is there some logic I need to look at in do_alert where something recovers and can't be ack'd again?
-Sean