Solaris

UNIX Friends,

Came in this morning to a downed E450 - thought it had been a power disaster. But wasn't! Machine had apparently shut itself down.

All three power supplies are double-green, and case interlock switch is closed...

As this is the first time we've had such a catastrophic(?) failure, am now taking the approach of removing components (with a bit of vacuum cleaning), and replacing one by one to see if one is defective. With CPUs, DC modules and memory removed, no combination of key position, power switch, etc. gets us back to the boot prompt. Having never tried this before, is this expected behavior?

Qs:
Is this diagnostic approach reasonable?
Will the machine get through its POST, and to a boot prompt, with no CPUs, no DC power modules, or memory?
If not, what is the minimum component content for the machine to power up?
Are there Best Practices Strategies for performing these (failed component) diagnostics?
Yes, with the machine up, in the past, a check of the log would reveal drive problems, overheating problems, etc.
But, with no Boot Prompt available, what's a lowly SysAdmin to do?

Many thanks for your wisdom and insights.

L

Sounds like you've checked this already, but I've had E450s that wouln't power up because the case wasn't seated properly in the past, so maybe double check that or when you have the cover off check the switch itself to make sure it's not broken.

In 10+ years working on Sun hardware I never ran into any other situation like you describe where the fans won't even spin up if the power supplies themselves were working. Do you have a different E450 that you could swap power supplies with and see if that helps? I wonder if they are dead in some way that the green lights are still staying on but they are actually failed.

rhfrommn - Thanks for that response. Yes, have just edited my post because I have now gotten powered up - indeed, one of the case switches was not fully closed. Have run into this myself before. Kind of a dummy error, I grant you...

Now it's more a question of: Powered Up, but no Boot Prompt. How to best diagnose what component might be at fault? (As I do hope it's just one component!)

get serial line connected and put the keyswitch to diag. after that run the POST.

How shall the POST be executed then, with no CPUs and no power?

On a 280R I had a similar problem. The cause was, that the CPU in slot 0 was defect and obviously prevented the machine from doing anything. I replaced it with the CPU from slot 1 and could get the machine up and running this way.

Hergp, thanks for your response.

(By way of clarification: I do, of course, have the power supplies still in the machine - they're all reporting good. I had been removing the CPU DC power modules as part of the diagnostic).

This machine did finally restart - and least got to POST. And it did so - interestingly - with no CPUs, memory or DC power modules.

Then, after lots of vacuuming and re-seating of components one-by-one, it appeared that one DC power module may have been the culprit. Another re-seat, and restart, and it's been running ever since. Has anyone ever heard of an intermittent problem with a DC power module? Should we be worried?

Still don't know what took it down, but we do have some evidence of a late-night power spike, so, am hoping that was the cause...

(No, no power conditioning on this machine for the moment - it's a testbed, so we're not too concerned with its real-world reliability for the moment.)