watch_in does not match on the name parameter as documented

[gthb]

With this in nginx/init.sls:

nginx-systemwide:
  pkg:
    - name: nginx
    - installed
  service:
    - running
    - name: nginx
    - reload: true

... and this in webapp/deploy.sls:

include:
  - nginx

{% for sitename, site in pillar.get('sites', {}).iteritems() %}
/etc/nginx/sites-available/{{ sitename }}:
  file.managed:
    - source: salt://webapp/site-nginx.jinja
    - context:
        site: {{ site }}
    - template: jinja
    - watch_in:
      - service: nginx
{% endfor %}

... an error is raised:

local:
    Data failed to compile:
----------
    Cannot extend ID nginx in "base:datamarket.webapp.deploy". It is not part of the high state.

But if I change the nginx declaration to have that exact ID:

nginx:
  pkg:
    - installed
  service:
    - running
    - reload: true

then it works. This is inconsistent with Requisites documentation which states “Requisites match on both the ID Declaration and the name parameter.” The requisite is not matching on the name parameter.

Worse (because it was harder to troubleshoot), the watch_in on service: nginx did not raise an error when I had that original nginx-systemwide state declaration and this other one:

{% for user, args in pillar['users'].iteritems() %}
{{ user }}:
  user:
    - present
    - name: {{ user }}
    - home: /home/{{ user }}
    - shell: {{ pillar.get('unprivileged_shell', '/bin/bash') }}
    - fullname: {{ args.get('fullname', user) }}
{% endfor %}

where, you guessed it, one of the users was named nginx. Of course this nginx didn't have any service state, so watch_in on service: nginx should have raised a clear error, something like “Can't apply watch_in to service: nginx because nginx has no service state”. But it didn't raise any error; it just silently failed to reload my nginx service when changing the file.

[basepi]

Thanks for the report. As far as I know, your first example should definitely be working.

Additionally, it definitely looks to me like we should not be silently failing on watch_in -- there may be a use case to explain why that silently fails, but offhand, we should fix that as well.

[basepi]

Can you please post your version info? salt --versions-report

[gthb]

$ salt --versions-report
           Salt: 0.17.2
         Python: 2.7.3 (default, Jan  2 2013, 13:56:14)
         Jinja2: 2.6
       M2Crypto: 0.21.1
 msgpack-python: 0.1.10
   msgpack-pure: Not Installed
       pycrypto: 2.6
         PyYAML: 3.10
          PyZMQ: 13.1.0
            ZMQ: 3.2.3

[ghost]

FWIW, I'm seeing this exact same issue.

$ salt --versions-report
Salt: 0.17.2
Python: 2.7.3 (default, Sep 26 2013, 20:03:06)
Jinja2: 2.6
M2Crypto: 0.21.1
msgpack-python: 0.1.10
msgpack-pure: Not Installed
pycrypto: 2.4.1
PyYAML: 3.10
PyZMQ: 13.0.0
ZMQ: 3.2.2

[terminalmage]

I'm able to reproduce this, and not just on watch_in, but require_in as well. All *_in requisites are disambiguated and turned into require/watch/etc. requisites when the state data is being compiled, so I have a good idea where this might be happening.