I rebooted my RedHat 7.3 server. It was having some kind of wacky file descriptor insanity. When it came back on-line I was unable to ssh in because the /etc/nologin was still present. I had to ssh in as root (only possible because I had overlooked this security hole) and remove the file before I could log in as a normal user. Expected Results: I think the mature UNIXes deal with this problem by performing an "rm /etc/nologin" somewhere in their init scripts. RedHat should do this too. rc.sysinit is the likely candidate.
This was determined as not the way to do it back in 1999; I'm not sure I want to change this behavior now.
So, are you saying that when the reboot malfunctions somehow and the machine boots with an /etc/nologin that the only resolution is physical access by someone with the root password? If this is the case, how do I prevent /etc/nologin from being created during the reboot process?
*** Bug 78129 has been marked as a duplicate of this bug. ***
The other UNIXes have this behavior (which has been in place for decades) correct. A reboot should delete the /etc/nologin file.
I'm happy to see that 8.5 years later, RHEL7 agreed with me. :)