[Bug 1582813] Re: service --status-all always reports upstart managed daemons as running

[Mathew Hodson]

** Description changed:

- [SRU Justification]
- SRU regression.  The change to lsb-base that landed into -updates on the 12th of May:
+ [Impact]
+ 
+ This is a regression caused by the change to lsb-base that landed in
+ -updates on the 12th of May:
  
  lsb (4.1+Debian11ubuntu6.1) trusty; urgency=medium
  
    * Add 01-upstart-lsb from the upstart package and make
      executing rc.d scripts no-op if there is an upstart
      job for that script. (LP: #1273462)
  
   -- Zhang Hua <joshua.zhang at canonical.com> Thu, 06 Aug 2015 11:11:16
  -0500
  
- changed the behaviour of init.d scripts to pass through to the
+ This changed the behaviour of init.d scripts to pass through to the
  associated upstart configuration for process management, where packages
  ship both an upstart configuration and an init.d configuration.
  
  This has a rather nasty side effect on 'service --status-all', which
  calls the 'status' method on all installed init.d scripts - this gets
  translated into:
  
    status <service-name>
  
  for upstart scripts, which correctly identifies the process as not
  running, but always returns a 0 return code. As a result, --status-all
  incorrectly identifies upstart processes as running even when they are
  not.
  
- [Regression potential]
- Minimal.  This SRU updates the behavior of the lsb init script upstart diversion to implement LSB-compliant return values for the 'status' subcommand.  It is possible this behavior change will differ from the original behavior of some init scripts, but that behavior was undefined.
+ [Test Case]
  
- [Test case]
  1. On an installed trusty system, run this command to get a list of affected upstart jobs:
- for job in $(initctl list | awk '/stop\/waiting/ { print $1 }'); do if [ -x "/etc/init.d/$job" ]; then echo $job; done
+ for job in $(initctl --system list | awk '/stop\/waiting/ {print $1}'); do if [ -x "/etc/init.d/$job" ]; then echo $job; fi; done
+ 
  2. Check the status of this service in the output of service --status-all:
  service --status-all 2>&1 | grep procps
+ 
  3. Confirm that the service is shown as running, despite us knowing above that it is not:
-  [ + ]  procps
+  [ + ]  procps
+ 
  4. Install the lsb-base package from trusty-proposed.
+ 
  5. Check the status of this service again in the output of service --status-all:
  service --status-all 2>&1 | grep procps
+ 
  6. Confirm that the service is no longer shown as running:
-  [ - ]  procps
+  [ - ]  procps
+ 
+ [Regression Potential]
+ 
+ Minimal. This SRU updates the behavior of the lsb init script upstart
+ diversion to implement LSB-compliant return values for the 'status'
+ subcommand. It is possible this behavior change will differ from the
+ original behavior of some init scripts, but that behavior was undefined.

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to lsb in Ubuntu.
https://bugs.launchpad.net/bugs/1582813

Title:
  service --status-all always reports upstart managed daemons as running

Status in lsb package in Ubuntu:
  Fix Released
Status in lsb source package in Trusty:
  Fix Committed

Bug description:
  [Impact]

  This is a regression caused by the change to lsb-base that landed in
  -updates on the 12th of May:

  lsb (4.1+Debian11ubuntu6.1) trusty; urgency=medium

    * Add 01-upstart-lsb from the upstart package and make
      executing rc.d scripts no-op if there is an upstart
      job for that script. (LP: #1273462)

   -- Zhang Hua <joshua.zhang at canonical.com> Thu, 06 Aug 2015 11:11:16
  -0500

  This changed the behaviour of init.d scripts to pass through to the
  associated upstart configuration for process management, where
  packages ship both an upstart configuration and an init.d
  configuration.

  This has a rather nasty side effect on 'service --status-all', which
  calls the 'status' method on all installed init.d scripts - this gets
  translated into:

    status <service-name>

  for upstart scripts, which correctly identifies the process as not
  running, but always returns a 0 return code. As a result, --status-all
  incorrectly identifies upstart processes as running even when they are
  not.

  [Test Case]

  1. On an installed trusty system, run this command to get a list of affected upstart jobs:
  for job in $(initctl --system list | awk '/stop\/waiting/ {print $1}'); do if [ -x "/etc/init.d/$job" ]; then echo $job; fi; done

  2. Check the status of this service in the output of service --status-all:
  service --status-all 2>&1 | grep procps

  3. Confirm that the service is shown as running, despite us knowing above that it is not:
   [ + ]  procps

  4. Install the lsb-base package from trusty-proposed.

  5. Check the status of this service again in the output of service --status-all:
  service --status-all 2>&1 | grep procps

  6. Confirm that the service is no longer shown as running:
   [ - ]  procps

  [Regression Potential]

  Minimal. This SRU updates the behavior of the lsb init script upstart
  diversion to implement LSB-compliant return values for the 'status'
  subcommand. It is possible this behavior change will differ from the
  original behavior of some init scripts, but that behavior was
  undefined.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/lsb/+bug/1582813/+subscriptions