Hm, after more analysis I'm no longer convinced that this is
appropriate.

In the openssh-server case, the if-up.d script just calls the initscript
reload action, which sends a SIGHUP to the process (if any). This
doesn't have any side-effects.

This is not the same for dhcp3-server. There is no reload action so we call 
restart, which does stop/sleep 2/start.
* this is called when interfaces are brought up at boot time, resulting in boot 
delay
* "/etc/init.d/dhcp3-server start" is called after the if-up.d script already 
started dhcp3-server, which is confusing, to say the least
* it results in DHCP server not serving addresses for a small time every time 
an interface is brought up
* More unanticipated breakage potential

The correct way to do it would be to have :
- an ifup.d script that returns immediately if dhcp3-server isn't already 
started, so that it doesn't affect the boot. However dhcp3-server doesn't 
really start a process if there are no interfaces available, so the resulting 
ifup script would not start it if needed interfaces are later brought up... 
defeating the whole point of this bug, I suppose
- a graceful restart feature in dhcp3-server so as to minimize those reloads 
impact

So my guess is that dhcp3-server isn't sufficiently smart to be
integrated into ifup/ifdown without dirty side-effects...

-- 
dhcp3-server needs if-up.d/if-down.d scripts for better network-manager 
compatibility
https://bugs.launchpad.net/bugs/280123
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to