Description of problem:
nrpe fails to start on boot with the following message in journal
Network server bind failure (99: Cannot assign requested address)
In /etc/nagios/nrpe.cfg I have set
because my host has several networks devices I do not want to bind to all of them.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Uncomment the server_address line from nrpe.cfg
2. Set nrpe to start on boot
nrpe fails to start
nrpe starts on boot
If I do 'systemctl start nrpe.service' after boot, nrpe starts
If I comment the 'server_address' line, then nrpe starts on boot
This package has changed ownership in the Fedora Package Database. Reassigning to the new owner of this component.
This sounds like your network connection that is assigning xxx.xxx.xxx.xxx is coming up late in the boot cycle, AFTER nrpe is attempting to start.
Yes, I agree
As such, there is no way for nrpe to bind to an address that doesn't exist, and I would have to mark this as expected behavior.
This problem reported 3 1/2 years ago, and yet, here i am, having my time wasted, because the problem still exists in el7, and fedora 27.
Closing this bug was a pathetic response.
How about setting up your initscripts or systemd units so nrpe starts AFTER the networking units eh?
If this is too much bother for *YOU* to do, what in the seven hells gives you the right to waste your user's time by requiring them to do it?
Until nagios maintainers address this issue, you can resolve it by creating a systemd drop-in. Create the file: /etc/systemd/system/nrpe.service.d/nrpe.conf containing:
Reboot, and you should see that systemd has started nrpe after your network has come up, and so nrpe should start successfully.
The systemd unit files (and initscripts for sysv distros) for nagios *should* do this themselves, and it is a bug in nrpe that they don't
Whoever looks after nrpe should fix this ASAP.