I have a CentOS7 server running named-chroot. The problem is named and named-chroot are separate systemd services in CentOS7, and I have both a stable system and a short memory. Months go by between any need for BIND maintenance. When I have BIND work to do, I always forget to type
systemctl restart named-chroot
and instead type
systemctl restart named
If I get it wrong, there's no warning that I'm starting the incorrect service.
I'd like to use systemd configurations to edit the named service to remind me that I've made a mistake. This is as far I've gotten to overriding the standard named service: I use
systemctl edit named
and put in the following:
[Unit] Description=You want named-chroot Wants= Before= After= [Service] Type=simple EnvironmentFile= Environment= ExecStartPre= ExecStart= ExecStart=/bin/bash -c "/usr/bin/echo 'No, you want named-chroot'" ExecReload= ExecReload=/bin/bash -c "/usr/bin/echo 'No, you want named-chroot'" ExecStop= ExecStop=/bin/bash -c "/usr/bin/echo 'No, you want named-chroot'" StandardOutput=journal+console PrivateTmp=false
This is as far as I got. The results of the echo commands will be visible in journalctl and systemctl status. Of course, the ideal would be to have them print on the terminal, but I haven't figured that out. This does not work:
nor variants thereof; I get error messages that tell me that /dev/tty is not available in bash!
Another problem: even with the above file, commands like these one can cause named-chroot to halt with no visible indicator or warning:
systemctl restart named systemctl start named
I can't figure out why the named daemon initiated by named-chroot should be affected by my revised named script, when I've overridden every named service option I could find.
If you're a systemd expert, please help me make this work.