* Don't do a separate "ntp -q -g" to reset the system time in the ntp
start script. It's probably not necessary, but more importantly, it can hang forever if the network is down. (Actually it will sit in a loop waiting for UDP packets that will never arrive.) This causes the NixOS reboot to hang, since Upstart can't kill jobs stuck in their start script. svn path=/nixos/branches/modular-nixos/; revision=15829
This commit is contained in:
		
							parent
							
								
									720d51179e
								
							
						
					
					
						commit
						674d63e212
					
				@ -93,11 +93,16 @@ mkIf config.services.ntp.enable {
 | 
			
		||||
            # Needed to run ntpd as an unprivileged user.
 | 
			
		||||
            ${modprobe}/sbin/modprobe capability || true
 | 
			
		||||
 | 
			
		||||
            ${ntp}/bin/ntpd -q -g ${ntpFlags}
 | 
			
		||||
            # !!! This can hang indefinitely if the network is down or
 | 
			
		||||
            # the servers are unreachable.  This is particularly bad
 | 
			
		||||
            # because Upstart cannot kill jobs stuck in the start
 | 
			
		||||
            # phase.  Thus a hanging ntpd job can block system
 | 
			
		||||
            # shutdown.
 | 
			
		||||
            # ${ntp}/bin/ntpd -q -g ${ntpFlags}
 | 
			
		||||
 | 
			
		||||
        end script
 | 
			
		||||
 | 
			
		||||
        respawn ${ntp}/bin/ntpd -n ${ntpFlags}
 | 
			
		||||
        respawn ${ntp}/bin/ntpd -g -n ${ntpFlags}
 | 
			
		||||
      '';
 | 
			
		||||
    }];
 | 
			
		||||
  };
 | 
			
		||||
 | 
			
		||||
		Loading…
	
	
			
			x
			
			
		
	
		Reference in New Issue
	
	Block a user