We have almost 20 Cybex “chassis,” rack-mounted enclosures that can contain a variety of cards that do things like connect to a computer, provide a user station, or network to another chassis. Since yesterday morning, almost none of the interconnectivity has been working. That’s what I’ve been working on nigh-exclusively since I arrived, then.
I just made some progress. I’ve isolated the problem.
It’s not Chassis 5, and it’s not Chassis 7/16 (a split chassis, thus with two identification numbers). It’s either Chassis 2 or Chassis 1 (which only connects to the rest of the network via #2).
Here’s how I got this far: I discovered that I could make a particular combination of inserted cards come up “green” on Chassis 5 (our main “routing hub” chassis), and that the trick to doing so was to leave out any card that connected in even a roundabout way to Chassis 7/16 (a secondary “routing hub” chassis). “Great, it must be in 7/16,” I thought.
In trying to reproduce this success on 7/16, I discovered that the only way to get reliable results was to leave out the card that connects the 7-side of that 4080 to Chassis 2. So I pulled the two cards in #2 that talk to the rest of the system, and reinserted all of the non-Enco cards in both 7/16 and 5. Voila! I can reliably pull up machines across all of the Server-room side of the Cybex network, and my XPRB (basically my “console unit” card) can display inventory for any connected chassis. At this point we have a mostly-usable system for the first time in a couple of days.
So now it’s just Chassis 1 and 2 that are isolated, and I need to figure out how to proceed from here… since isolating the problem doesn’t explain the problem itself.
Hoo boy. At least I can say I did something valuable with my time today, eh?
Comments
One response to “Progress with Cybex”
The process was absolutely correct. A single XPAC or similar card can cause a Cybex XP chassis to not appear. Basically, make the entire chassis unaccessible. The trick to finding the bad card – resolving the issue follows your original approach.
Connected the chassis to your main system with a single XPST connection. Remove all oteher cards except one operational server card (XPAC or XPAB). Verify a connection to that server. Then, start seating the other server cards one at a time validating each time connectivity is still available.
When the connection is loss, the last seated card is bad, unseat it and proceed to check the other cards. Technically, the chassis should be powered down each time a card is seated.