[MIR] insserv

Bug #427709 reported by Steve Kowalik
28
This bug affects 3 people
Affects Status Importance Assigned to Milestone
insserv (Ubuntu)
Fix Released
Critical
Loïc Minier
Karmic
Fix Released
Critical
Loïc Minier

Bug Description

Binary package hint: sysvinit

With the latest upload of sysvinit, update-rc.d assumes insserv is available, and attempts to call it if the init script exists. This assumption doesn't hold for Ubuntu system, since we don't use insserv.

From a cleanly debootstrap'd karmic chroot:

root@thimbleberry:/# update-rc.d module-init-tools start 15 S .
perl: warning: Setting locale failed.
perl: warning: Please check that your locale settings:
 LANGUAGE = (unset),
 LC_ALL = (unset),
 LANG = "en_AU.UTF-8"
    are supported and installed on your system.
perl: warning: Falling back to the standard locale ("C").
update-rc.d: using dependency based boot sequencing
Can't exec "insserv": No such file or directory at /usr/sbin/update-rc.d line 402.

The command is taken from module-init-tools' postinst, and due to that call failing, Karmic can not be deboostrap'd.

Tags: ubuntu-boot
Steve Kowalik (stevenk)
Changed in sysvinit (Ubuntu):
importance: Undecided → High
Revision history for this message
Noel J. Bergman (noeljb) wrote :

I saw insserv install this morning when doing updates:

...
2009-09-10 11:23:52 upgrade lsb-base 4.0-0ubuntu3 4.0-0ubuntu4
2009-09-10 11:23:57 upgrade lsb-release 4.0-0ubuntu3 4.0-0ubuntu4
2009-09-10 11:23:58 upgrade initscripts 2.86.ds1-61ubuntu16 2.87dsf-4ubuntu1
2009-09-10 11:23:58 upgrade sysv-rc 2.86.ds1-61ubuntu16 2.87dsf-4ubuntu1
2009-09-10 11:23:58 install insserv <none> 1.12.0-11
...

It appears that sysvinit depends on sysv-rc which depends on insserv. There appears to be a corresponding comment in the changelog:

  * Move the code to migrate to dependency based boot sequencing
    during upgrades from the insserv package to the sysv-rc package.
    Depend on insserv (>> 1.12.0-10) for this. Let initscripts depend
    on sysv-rc | file-rc to make sure they are installed first.
    Migration is a one-way process, enabled after a critical debconf
    question during upgrades when it is safe to do so (Closes:
    #540546, #541041, 541604). Checks previously done by
    update-bootsystem-insserv are now only done once in sysv-rc
    postinst (Closes: #538934). Dependency based boot sequencing is
    now the default. This change make it possible to remove both
    sysv-rc and insserv (Closes: #538959) if other packages want to
    take over the boot sequencing resposibility.

Revision history for this message
Scott James Remnant (Canonical) (canonical-scott) wrote : Re: [Bug 427709] [NEW] update-rc.d assumes insserv is available

On Fri, 2009-09-11 at 06:01 +0000, Steve Kowalik wrote:

> With the latest upload of sysvinit, update-rc.d assumes insserv is
> available, and attempts to call it if the init script exists. This
> assumption doesn't hold for Ubuntu system, since we don't use insserv.
>
Since it *DEPENDS* on it, it's _allowed_ to assume it's available.

 status invalid

Scott
--
Scott James Remnant
<email address hidden>

Changed in sysvinit (Ubuntu):
status: New → Invalid
Revision history for this message
Loïc Minier (lool) wrote : Re: update-rc.d assumes insserv is available

@Scott: StevenK told me this morning that this was the bug breaking today's livefs builds e.g.:
http://people.canonical.com/~ubuntu-archive/livefs-build-logs/karmic/ubuntu-netbook-remix/latest/livecd-20090911-i386.out

Could you reassign instead of invalidating if you think it's not a sysvinit issue?

Revision history for this message
Randall Leeds (randall-leeds) wrote :

This bug appears to make it impossible right now to produce a karmic chroot.
Needless to say, big problem.

I'm going to change this back to valid, but only after changing the description to be maybe more accurate.
Please correct me if I'm wrong.

summary: - update-rc.d assumes insserv is available
+ sysv-rc depends on insserv but it is not installable
Changed in sysvinit (Ubuntu Karmic):
status: Invalid → New
status: New → Invalid
status: Invalid → New
Revision history for this message
Colin Watson (cjwatson) wrote : Re: sysv-rc depends on insserv but it is not installable

Should we turn this bug into an MIR for insserv, then?

Revision history for this message
Loïc Minier (lool) wrote :

I asked cjwatson to please prepromote this as it breaks builds

affects: sysvinit (Ubuntu Karmic) → insserv (Ubuntu Karmic)
Changed in insserv (Ubuntu Karmic):
importance: High → Critical
milestone: none → karmic-alpha-6
status: New → Triaged
summary: - sysv-rc depends on insserv but it is not installable
+ [MIR] insserv
Revision history for this message
Colin Watson (cjwatson) wrote :

I have promoted insserv for the time being, at Loïc's request.

Revision history for this message
Scott James Remnant (Canonical) (canonical-scott) wrote : Re: [Bug 427709] Re: sysv-rc depends on insserv but it is not installable

On Fri, 2009-09-11 at 15:37 +0000, Colin Watson wrote:

> Should we turn this bug into an MIR for insserv, then?
>
Argh, sorry, that's my bad.

It shows up as in main for me already, but I've just realised that's
because *everything* from a PPA shows up as in main.

Yes, please promote to main.

Scott
--
Scott James Remnant
<email address hidden>

tags: added: ubuntu-boot
Revision history for this message
Steve Langasek (vorlon) wrote :

Not a blocker for alpha 6, shuffling the milestone. Nevertheless this MIR should be finished ASAP.

Changed in insserv (Ubuntu Karmic):
milestone: karmic-alpha-6 → ubuntu-9.10-beta
Revision history for this message
Scott James Remnant (Canonical) (canonical-scott) wrote :

 1. Availability: The package is in Ubuntu and built for all architectures

 2. Rationale:
    - the package is a dependency of sysv-rc
    - while we won't actively use it:
       - Debian are actively using it, and it creates good will to have it available to Ubuntu users as well in the "can switch on" state
       - I'm not sure that sysv-rc works anymore without it at least existing on the system ;)

 3. Security: No known advisories or problems.

 4. QA: maintained by Debian sysvinit maintainers, part of lenny

 5. UI standards: N/A

 6. Standards compliance: lots of this

 7. Dependencies: all in main

 8. Maintenance: maintained in Debian, this is a sync only

Changed in insserv (Ubuntu Karmic):
assignee: nobody → Robbie Williamson (robbie.w)
Changed in insserv (Ubuntu Karmic):
assignee: Robbie Williamson (robbie.w) → MIR approval team (ubuntu-mir)
Revision history for this message
Loïc Minier (lool) wrote :

So my understanding is that even if we transition everything in Ubuntu main to upstart, we will keep sysv-rc compat which needs insserv as implemented in Debian. That sounds ok but it would be nice if you could subscribe to its bugmail in Ubuntu (not a lot it seems).

Low number of bugs in Debian.

Maintenance in Debian looks solid, packaging is in decent shape and up-to-date

Approved

Changed in insserv (Ubuntu Karmic):
assignee: MIR approval team (ubuntu-mir) → Loïc Minier (lool)
status: Triaged → Fix Committed
Revision history for this message
Martin Pitt (pitti) wrote :

This was pre-promoted, closing.

Changed in insserv (Ubuntu Karmic):
status: Fix Committed → Fix Released
Revision history for this message
Scott James Remnant (Canonical) (canonical-scott) wrote : Re: [Bug 427709] Re: [MIR] insserv

On Thu, 2009-09-17 at 21:20 +0000, Loïc Minier wrote:

> So my understanding is that even if we transition everything in Ubuntu
> main to upstart, we will keep sysv-rc compat which needs insserv as
> implemented in Debian.
>
Initially, yes.

A longer term plan is to actually have Upstart parse LSB-compatible
files from /etc/init.d itself - but that's far enough away that we need
the interim solution

> That sounds ok but it would be nice if you could
> subscribe to its bugmail in Ubuntu (not a lot it seems).
>
I am? (I just double-checked, it says I'm subscribed :p)

Scott
--
Scott James Remnant
<email address hidden>

Revision history for this message
Loïc Minier (lool) wrote :

Odd, I checked about one hour before making the comment yesterday evening; perhaps I looked at the wrong package but that would be odd

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.