On Wed, Jan 10, 2007 at 02:26:46PM -0600, Narayan Desai wrote: > Sami> I agree here, a simple failback to force-reload would be a > Sami> simple solution. But on a long term it would be a good idea to > Sami> catch the errors and report them back to the server and > Sami> possibly attempt reconfigure later. > > Should we change the default to force-reload for debinit? it can be > overridden by using the reload attribute for services. If force-reload > is mandatory, then we should probably use it. Thoughts?
That sounds like dangerous. Some daemons really do not like force-reload. Autofs for example will try to unmount everything on a force-reload with bad consequences we users are logged in and their homes mounted. Now, setting and option to use reload instead of force-reload would require someone to look at every service entry in the repository before doing the upgrade. That does not sound like an easy upgrade. I think that the fallback to force-reload when the script returns an usage line is more user friendly. > We have an analogous issue with fedora, except force-reload -> > condrestart, so I was thinking about changing it chkconfig to use > condrestart... I don't know fedora but I guess that the same applies. thanks jacques
signature.asc
Description: Digital signature
