05-21-2012 07:55 AM
We're opening a case with TAC now, but I wanted to check if anyone has seen controllers randomly rebooting with "Nanny rebooted machine - cfgm process died."
We just upgraded from 18.104.22.168 to 22.214.171.124 last night and several of our 651s are rebooting with that cause. We didn't go to 126.96.36.199 because I observed odd VRRP behavior with that release in a prior upgrade in a different infrastructure. However, there's nothing to indicate that anything like this is fixed in 188.8.131.52 anyway.
(MIA-WLC01) #show ver
Aruba Operating System Software.
ArubaOS (MODEL: Aruba651-US), Version 184.108.40.206
Copyright (c) 2002-2012, Aruba Networks, Inc.
Compiled on 2012-01-25 at 23:06:42 PST (build 32089) by p4build
ROM: System Bootstrap, Version CPBoot 220.127.116.11 (build 23274)
Built: 2010-01-19 11:11:41
Built by: p4build@re_client_23274
Switch uptime is 30 minutes 44 seconds
Reboot Cause: Nanny rebooted machine - cfgm process died.
Processor XLS 408 (revision B1) with 906M bytes of memory.
32K bytes of non-volatile configuration memory.
256M bytes of Supervisor Card System flash (model=NAND 256MB).
Solved! Go to Solution.
05-21-2012 08:15 AM
Do you utilize the internal AP on the 651? If not, can you disable it? I think that will stop the reboots until Aruba fixes the underlying bug.
If you do use the internal AP, you may have to downgrade.
05-21-2012 11:24 AM
You need to make a new AP group and the remove all of the VAPs. Also, make a two new radio profiles and make sure the radio's are disabled.
Then, reprovision the internal 651 AP into that new AP group so that it will be disabled.
Any other APs that use the 651 should stay in the groups they are already provisioned into.
Make sure you don't disable the radios on the production radio profiles or your APs will stop working.
05-21-2012 11:34 AM
Ok, that makes sense, thanks. One point of clarity, I have about five of these and I'd rather just make one AP group for this workaround. Is it at all relevant what LMS is in the AP system profile? I was just going to use my master controller for that if it's ok.
Before we do anything though, we're going to get final word from TAC.
Honestly, I'm flat out shocked that Aruba would allow this code to even be downloadable. You simply can't put code out for general use that is knowingly going to crash a platform in just about every single out-of-the box use case. There at least needs to be a warning in the download directory that the code is not acceptable for use in a 651.
05-21-2012 11:39 AM
Definitely have TAC confirm that disabling the radio will help.
You can create the AP group and related profiles from the master and have it push to the locals. That's not a problem.
05-21-2012 11:43 AM
Yes, I know I can create the group on the master, I was just wondering if it would be ok to use the master as the LMS for all five of the 651's built-in APs. I don't see why it would be a problem, but I wanted to check.