kernel panic - not syncing: Attempted to kill inint !

[DEBCEN]

Hi,
I've purchased a dedicated server today(CentOS 6.7) and it came SELinux disabled by default. I turned it to enforcing mode and changed the policy type to strict(like I had in my previous server). I rebooted the server and I no longer can connect through SSH. I connected through IPMI and got this message at boot :

Code: Select all

kernel panic - not syncing: Attempted to kill inint !
PId: 1, comm: init not tainted 2.6.32.-279-5.2.e16.x86_64 #1
Call Trace:
[<ffffffff814fd39a>] ? panic+0xa0/0a0x168
[<ffffffff81070bd2>] ? do_exit+0x862/0x870
[<ffffffff8117ccf5>] ? fput+x25/0x30
[<ffffffff81070c38>] ? do_group_exit+0x58/0xd0
[<ffffffff81070cc7>] ? sys_exit_group+0x70/0x20
[<ffffffff8100b0f2>] ? system_call_fastpath+0x16/0x1b

The only advice I could get was to disable SELinux but I can't figure out how to do it since I have no physical access to the server. Why is this happening as I have this configuration running perfectly for my previous server. Is there a way out of this problem other that OS reload?
Any help is appreciated.
Thank You

[TrevorH]

You have access to your server via IPMI so you can use that to reboot and make changes. Start by editing the kernel command line and adding enforcing=0 to it so that it boots in permissive mode. Then touch /.autorelabel and edit the /etc/sysconfig/selinux file to set it to start in permissive mode and then reboot. This boot will take longer as it will relabel all files with their correct contexts.

changed the policy type to strict

What exactly did you do?

[DEBCEN]

Thank you, I'm now working on it. Will keep this thread posted.

What exactly did you do?

I don't remember what I did but I changed something that was by default "targeted" to "strict"
Thanks for the reply

[TrevorH]

Change it back to targeted

[DEBCEN]

I spent nearly a day to get this problem fixed. The OS don't boot at all, just showing the above mentioned error when it boots. Finally, I live booted the rescue system, and disabled SELinux by placing selinux=0 on grub.conf as I could not see any other directory like /etc of my CentOS in it. Now it is working fine but the thought that SELinux is disabled on my system is uncomfortable. Any idea why this happened ? If I re-enable it with type targeted, are you sure that it will not happen ?(just don't want to spent another 10+ hours of effort)
Thank you for your reply.

[TrevorH]

Did you follow my post to your original? The second one in this thread?

You didn't need to boot a rescue media and edit grub.conf - you can do that from the IPMI console and interrupt it at the grub menu and use 'e' to edit the entry and append 'enforcing=0' to the command line. Now boot in permissive mode, create the /.autorelabel file and reboot and let it relabel the entire filesystem.

[DEBCEN]

you can do that from the IPMI console and interrupt it at the grub menu and use 'e' to edit the entry and append 'enforcing=0' to the command line

Thanks to that advice, I tried that and it might had saved my day unfortunately, hitting the 'e' key continuously didn't open the command line on the console. When I start remote console, it opens the cosole blank for about 10 minutes and shows that error, I even tried hitting 'e' key continuously for 10 mins(till the error appeared). I opened a ticket with them and they told me to use rescue mode instead. The current situation is like this :
1)grub.conf restored to original after I successfully booted into the system
2)SELinux has been disabled
Now I need to re-enable SELinux and don't want that problem to happen again. Is it possible to enable SELinux without taking this risk ?
Thank you for helping me out

[TrevorH]

You're trying too late. The point you need to hit the key is when it displays the "Booting CentOS 6 2.6.32-whatever in X seconds"

[DEBCEN]

You're trying too late. The point you need to hit the key is when it displays the "Booting CentOS 6 2.6.32-whatever in X seconds"

When I was with my previous provider it worked fine but my new provider allows IPMI access only through their VPN.Now, when I IPMI boot the server, the console opens up and it just shows black(with no cursor nor anything) for about 10 minutes and all of a sudden the error shows -- I can't see anything like "Booting CentOS 6 2.6.32-whatever in X seconds", just the error.
I'll try to figure this out, whether it is my problem or theirs.
Thanks for helping me out, everything is *somewhat* fine now.