Recently, some of our hosts that are selinux enforcing are not able to run the check_raid nrpe check.
For example:
[kevin@noc01 ~]$ /usr/lib/nagios/plugins/check_nrpe -H virthost02.phx2.fedoraproject.org -c check_raid NRPE: Unable to read output
This may require changes to our selinux nrpe policy or the like. You can view selinux audit messages for the hosts on log02 and then adjust the nrpe selinux module in puppet.
Selinux issue fixed, but we still need to check why nrpe is unable to read the command output.
check_raid seems working as npre user
There is still a selinux issue it seems...
type=1400 audit(1323539541.859:879999): avc: denied { search } for pid=2066 comm="sh" name="home" dev=dm-1 ino=392456 scontext=unconfined_u:system_r:nrpe_t:s0 tcontext=system_u:object_r:home_root_t:s0 tclass=dir type=1400 audit(1323539541.892:880000): avc: denied { read } for pid=2066 comm="python" name="mdstat" dev=proc ino=4026531961 scontext=unconfined_u:system_r:nrpe_t:s0 tcontext=system_u:object_r:proc_mdstat_t:s0 tclass=file
Not sure why it's looking at home. ;(
Looking at this again, I think it is solved. it's showing the unable, but thats ok on those hosts because there's no /proc/mdstat (they are virtuals).
Closing this out now.