From: Gilles (gilles_at_harfang.homelinux.org)
Date: Sat 23 Apr 2005 - 12:30:53 BST
> > I'm trying to install a package in a Debian vserver,
> > but the configuration script ends with
> > kill -s HUP 1
> runit does that, this used to put this package in 'uninstallable' state,
That's the one indeed.
> because author assumed that kill -HUP 1 always succeeds.
> This is a mistake on packager's side. You can play with fakeinit to solve
> the situation, but in general software author should be educated.
Actually, I sent a bug report on March 12 about that script's behaviour.
But I didn't get any reaction.
I also sent a mail to their mailing list a few days ago...
> (and it's not like it's vserver-only issue, there are other legitimate
> reasons why 'kill 1' would fail).
That's why I hadn't asked here: At first I didn't notice the "kill" failure
because it happens only when "/etc/inittab" is unmodified; every time afterwards,
when "dpkg" tries to configure, it's another error that shows up. [It seems that
the "idempotency" rule (from the Debian Policy) isn't followed here.]
Then I thought about making the script happy anyway, with "fakeinit".
> Free software used to be about quality, and if you raised such issue to
> software author she used to happily fix it... not always the case now.
I just hope that it doesn't tell about the quality of "runit" itself...
I think that service supervision is a nice idea that fits well with vserver
(robustness and security).
Vserver mailing list