K2000 Freezing every couple of days...crashing PXE/TFTP boot
Our K2000 seems to be freezing/crashing every couple of days lately. All 5,000 of our machines are set to PXE boot first, so they are getting TFTP errors upon boot, and can't get into their machines (trying to tell them to hit F12 is pointless...they won't listen).
Anyone have any ideas on this? We are having to hard restart the box, but by the time the techs get in, over 2,000 of our users are at work already and can't boot.
This has happened randomly since we got the box, but seems to be more frequent lately.
Answers (4)
What is the console saying?
I had to disable C states on mine.
Comments:
-
I had the same problem, it was occuring monthlyish if memory serves me. I talked to support and they had me make some bios changes to the C states. - darkhawktman 11 years ago
-
I think it says something about "spinlock being help too long". - dugullett 11 years ago
-
Yes, please let us know the error in the konsole. "Spinlock being held too long" means that C states need to be disabled. A page fault 12 is a different error.
Corey
Lead L3 Enterprise Solutions Engineer, K2000
If my response was helpful, please rate it! - cserrins 11 years ago -
Thanks - I spoke with our networking team, and it seems that we did the C States BIOS change already in the past. I guess we'll wait to see what the error says next time. We're usually in a rush to get it back up from screaming staff to really pay attention I guess. - sfigg 11 years ago
We've experienced the same off and on and support was never able to nail down a cause. Of course, we never had any errors either. It would always freeze on the login screen at the console. I'm going to check to see if C States is on in the bios. Is there any recommended bios updates or versions for the K2?
Hello,
only an an idea: maybe that replying to a peak of 5000 client that are turning on around the same time is a bit ''destabilizing''...
I think that is better to grab all the log of the K2000 and open a call to the support so they may spot immediately something in the logs and/or conduct more analysys.
Another good information to send to the support is a Wireshark capture of the traffic when this condition happens.
Kind regards,
StockTrader
Comments:
-
Thanks for the idea - It's usually already down by the time people get in the building, so it crashed sometime after hours. - sfigg 11 years ago
