On Tue, 1 Feb 2005, Steve Traylen wrote:
> On Tue, Feb 01, 2005 at 09:58:11AM -0000 or thereabouts, Lev Shamardin wrote:
> > On Tue, Feb 01, 2005 at 09:23:19AM +0100, Oliver Keeble wrote:
> > > > Well, there is a script in the /etc/cron.d and it even works, but there are
> > > > two things that I can't really understand:
> > > >
> > > > 1. config_mkgridmap script does makeing the gridmapfile after the
> > > > reconfiguration. The config_crl does not fetch the CRLs. One could expect
> > > > quite the same behaviour from script from the same set.
> > >
> > > To clarify - are you saying that your configuration is in fact
> > > working, but you feel that the config_crl function should download the crl
> > > list when run? I agree with this point and will update the script.
> >
> > The configuration works. I'm saying that config_crl and config_mkgridmap
> > should have the same behaviour: ether both of them should run the cron job
> > that they schedule (update the gridmapfile & fetch the crl) or both of them
> > should not run the cron job. When one of the scripts does and another does
> > not it is quite confusing.
>
> I would say they should not be run at all, I think it is generally a good idea
> to not have nothing in the installation that depends on anything external
> off site. You run the risk that installations can be different to one another
> because something happening that you can't control. We obviously mirror all
> of your repositories to achieve this.
> Steve
Yeah, point taken, but such things would cause two otherwise
identical installations to diverge subsequently anyway. Clearly this is
undesirable, but I think it means you gain more than you lose having the
node as up to date as possible.
-Oliver
--------------------------------------------------------------------
Oliver Keeble Information Technology Department
[log in to unmask] CERN
+41 22 76 74929 CH-1211 Geneva 23
