equivalent of chkconfig

Tom H tomh0665 at gmail.com
Tue Aug 18 14:21:17 UTC 2015

On Tue, Aug 18, 2015 at 7:00 AM, Martin Pitt <martin.pitt at ubuntu.com> wrote:
> Tom H [2015-08-18 4:49 -0400]:

>>> update-rc.d and invoke-rc.d are tools for package maintainers only (to
>>> be used from pre/postinst scripts) and only applies to sysv-init
>>> scripts ...
>> Yes and no. They are meant for maintainer scripts but update-rc.d is
>> needed by admins for enabling/disabling daemons because there isn't
>> another tool available in Debian and Ubuntu.
> Right, and it's the very tool to do just that. What gives the
> impression that it is a maintainer scipt-only tool?

You can find many statements by Debian developers to that effect.

I suspect that it was due to its complex syntax pre-insserv:

update-rc.d daemon start 80 2 3 4 5 . stop 20 0 1 6 .

Why would Debian have created "service" if not because it didn't want
its users to use "invoke-rc.d"?

>> There was a Debian bug for service, the admin equivalent of
>> invoke-rc.d, to be enhanced to forward "service daemon enable|disable"
>> to update-rc.d in order to provide an admin interface but the Debian
>> systemd and upstart maintainers requested that this work with their
>> respective toys and the bug went nowhere.
> That would make things even more confusing IMHO. "service" is for
> runtime starting/stopping, "update-rc.d" for configuring which
> services start at boot. It has worked like that forever in
> Debian/Ubuntu, with any init system.

As a mostly RHEL admin until recently, I agree with you because

1) It would've made more sense to have a separate executable, perhaps
by adding (possibly modifying) chkconfig to the sysv-rc package (it's
packaged in Debian; it's not currently packaged in Ubuntu).

2) It would've meant that service on Ubuntu would've behaved
differently that on RHEL (and Solaris of old and possibly others),
although it would make sense to start/stop/enable/disable a daemon
with the same base command.

But this was discussed on debian-devel@ and there was an RFE bug
report opened for it.

More information about the Ubuntu-devel-discuss mailing list