Red Hat Bugzilla – Bug 163784
/etc/sysctl.conf not read during boot
Last modified: 2014-03-16 22:55:06 EDT
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.7.8) Gecko/20050513 Fedora/1.0.4-1.3.1 Firefox/1.0.4
Description of problem:
Added a value for kernel.printk in /etc/sysctl.conf. Value is not loaded during boot (via /etc/rc.d/init.d/network) but can be successfully loaded after boot using sysctl.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Set a custom value for kernel.printk in /etc/sysctl.conf
3. Verify value not set using cat /proc/sys/kernel/printk
4. Verify valid entry in /etc/sysctl.conf using sysctl -e -p /etc/sysctl.conf
5. Verify value is set using cat /proc/sys/kernel/printk
Did not find any related entries in /var/log/audit/audit.log. Upon further checking found auditd does not start until after network. Given that auditd only logs to a local file, is there any reason it should not be started before all else in runlevel 2?
Why did you assign this to SELinux policy? Shouldn't it be a initscripts
problem? Are you seeing SELinux failures on the console?
/var/log/audit may not be a local directory. I start it at the first opportunity
that its safe to do so. Also, chkconfig does not allow you to specify the start
order in a level by level way. If kernel cannot log to auditd, the message will
go to syslog. So, there is a safety net.
If you turn off SE Linux & reboot, does the parameter get loaded? Or maybe you
can do setenforce 0, service network restart, and then check for it. This avoids
needing to relabel the file system.
If not, this looks like initscript problem. Not audit or policy.
Transferring to initscripts this is not an audit problem.
The value that's exported through kernel.printk is reset by syslogd/klogd, which
runs after network is started.