Powering down a failed controller

[strider78]

Hello all!

We have a V6220 single-chassis dual-controller filer (controllers are named Filer-A and Filer-B). Some time ago Filer-A had a hardware failure (maybe voltage regulators but I'm not 100% sure), was taken over by Filer-B, rebooted by watchdog, but didn't came up. Now it's in zombie state, dead by fact but still powered on. Whole system works through alive Filer-B.

Is it safe to issue "system power off" command from Filer-A Service Processor CLI? Sorry for such a stupid question, but I'm not an expert in storage and we have a very mission-critical database on this system.

[aborzenkov · Accepted solution]

What is the reason to power it off? It is not required for replacement. I am not sure what "safe" means in this case. It should not affect another node in the same chassis if that was the question.

[strider78]

The reason is simple, it's a last resort, I hope that power cycle may heal it. Yes, I wanted to make sure that powering it off won't harm a working filer. Thank you very much for an answer.

[cedric_renauld]

HEllo, 

I think your system is in Wauting for give back or maybe @loader prompt

Can you type before :

system console 

And check the state of your controler 

Anf on your survival controler, wath is the result of

cf status 

Thanks  

[strider78]

Filer-A was dead completely, "system console" showed nothing, "cf status" on a survived Filer-B showed only that it has taken over a partner. Interconnect link was down too.

Then I issued "system power off" on a dead filer's SP. Powered it on after 5 minutes.

Bingo!

All hardware (voltage) issues were auto-deasserted, ONTAP booted normally and now is ready for a giveback.

Now I'm sure that my system was hit by a 500 days uptime bug. Upgraded my SP's firmware to latest compatible version.

Thanks all for help.