On 13/11/11 17:06, Andreas Stempfhuber wrote:
> Hi Simon,

> Regarding bug #506734, I have concerns because it creates an unexpected 
> shutdown behavior. It assumes that the shutdown order doesn't matter. But why 
> is there a Required-Stop option defined by LSB and why does dnsmasq use it to 
> define shutdown dependencies, if they don't matter? It seems to me that this 
> has not been considered by the discussion.

That's a very good question: the Required-Stop: dependencies have been
the same since the dnsmasq script was LSB-ised five years ago.

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=374650

has the gory details from back then. I actually don't think that the set
of dependencies was thought through very much: $network is probably not
needed for stopping, and lack of $syslog just loses the final
"Exiting...." message with current code, as we've discovered. I'm
worried about $remote_fs though; why could /usr be needed and would it
have gone away by sendsigs time?

http://wiki.debian.org/LSBInitScripts says:
        
 $remote_fs
        
all filesystems are mounted. In some LSB run-time environments,
filesystems such as /usr may be remote. If the script need a mounted
/usr/, it needs to depend on $remote_fs. Scripts depending on $remote_fs
do not need to depend on $local_fs. During shutdown, scripts that need
to run before sendsigs kills all processes should depend on $remote_fs.


As the dnsmasq binary is /usr/sbin/dnsmasq, I can't see that /usr would
be unmounted before the process dies.



> 
> The reason for bug #50673 was a faster shutdown by stopping the daemons in 
> parallel by sendsigs instead of running each shutdown script in serial. In 
> the 
> meantime Debian changed the init system and is executing the shutdown scripts 
> in parallel, one script can no longer simply delay all others. The reason for 
> bug #50673 (faster shutdown) does IMHO no longer exist.

The meta-reason for 50673 is that this change unified the Debian and
Ubuntu packages for dnsmasq. Given that Ubuntu has changed its init
system since then, it may be possible make the change back to stopping
at levels 0 1 and 6 without causing Ubuntu problems.

> 
> Independent of my opinion regarding bug #506734, dnsmasq is a great tool. It 
> is since years my first choice when I need a DNS and/or DHCP server. Many 
> thanks for this great software!

Glad you like it!


Cheers,

Simon.




-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org

Reply via email to