Re: Bug#904558: What should happen when maintscripts fail to restart a service
Apologies for the long delay.
We discussed this issue in several TC meetings without being able to
make
real progress.
After several rounds of discussions we came to the conclusion that the
reason why we can't make progress is that we always end up hitting the
wall
of "The Technical Committee does not engage in design of new proposals
and
policies". While we recognize that this is a problem worth fixing, this
is
not something that we can fix as a body and need the help of the
Developers
to do it.
On the one hand, maintainers want to be able to notify sysadmins when
things don't go as expected. On the other hand, sysadmins don't want
their
systems to be left in weird/broken states because one single thing
didn't
go as expected.
A failing maintscript is a horrible way of notifying sysadmins, but it's
the only one available up to now and so package maintainers use it when
they think the failure is critical enough.
So, the TC declines to rule on what should maintscripts do when failing
to
(re)start a service (or otherwise encountering a similarly serious
problem).
Instead, we recommend that a work group of developers is formed, to
create
a better mechanism of notification that can be used to let sysadmins
know
when things don't go as expected on their systems, without leaving the
machines in weird/broken states. Given that this is a problem faced by
many
Linux distributions, it would be nice if this mechanism was developed
and
published in a non Debian specific way that made it also available for
other
distributions to use.
Once that mechanism exists, we would strongly recommend that almost all
failures use this mechanism, instead of failing maintscripts.
--
Marga, on behalf of the Technical Committee
Reply to: