Hi all,
You have stumbled on a piece of deficiency that I have grumbled about before. Not only do stackables not tell you when a stack member does not come up from a power down/reset, chassis have the same issue. Worse, they forget to tell you about failed power supplies and failed fans. I got Extreme to put in traps for failing blades/stack members in newer code, so maybe with a little push and shove, we as a community can get them to do something a little smarter at reboot/restart time.
I would propose that a management module (on any type of device, red or purple) scan its hardware looking for objects that are configured but not operational. These objects should include fans, power supplies, blades, stack units, stacking cables, and anything else that only produces a single trap/log entry when that object fails. (Note that this is not a definitive list, feel free to add.)
Upon finding such (apparently) broken objects, send an alert (trap/log/syslog) message out indicating same. The gotcha here is you might need to do this a couple of times spaced say a minute apart in order to have a reasonable chance of getting through to your favorite network management system. (The uplinks need to be online to get anywhere!)
Reasonable?
James