foreman_config_entry consuming polluted value

[nbarrientos]

foreman_config_entry consumes the whole string printed to stdout by foreman-rake -- config -k foo when invoked.

If the apipie cache is not generated, Foreman prints to stdout some text, example:

# foreman-rake -- config -k rss_enable
API controllers newer than Apipie cache! Run apipie:cache rake task to regenerate cache.
false

So, as the requested value via foreman_config_entry does not match the current value, (it does actually, but it's prefixed by the apipie bit) then the setting is incorrectly re-set (making the Puppet run slower and adding rubbish to the audit log, for instance).

puppet-agent[348248]: (/Stage[main]/.../Foreman_config_entry[create_new_host_when_report_is_uploaded]/value) value changed "API controllers newer than Apipie cache! Run apipie:cache rake task to regenerate cache.\nfalse" to 'false' (corrective)

We have this patch locally:

code/lib/puppet/provider/foreman_config_entry/cli.rb | 1 +

modified   code/lib/puppet/provider/foreman_config_entry/cli.rb
@@ -17,6 +17,7 @@ Puppet::Type.type(:foreman_config_entry).provide(:cli) do
           :gid                => 'foreman' }.merge(options)
       )
       status = $?
+      output.slice! "API controllers newer than Apipie cache! Run apipie:cache rake task to regenerate cache.\n"
       output if status.success?
     end
   end

To prevent this from happening but of course it's not the best solution. Ideally the output produced by Foreman should go to stderr instead.

What do you think?

[ekohl]

To prevent this from happening but of course it's not the best solution. Ideally the output produced by Foreman should go to stderr instead.

Agreed. That particular line is here:
https://github.com/theforeman/foreman/blob/e2918f95f2c11bef3a287f63949387d47c771202/config/initializers/apipie.rb#L92

I'd happily support a patch that fixes it.

[nbarrientos]

Thanks. So just to confirm the next steps would be to close this issue, create one in the Foreman bug tracker and eventually send a patch for it, right?

[ekohl]

Yes, I think so. I'd also add a link to the Foreman Redmine issue here and refer to this issue in the Redmine issue just to keep it connected, but it wouldn't surprise me if you'd already do that anyway :)

[nbarrientos]

As discussed: https://projects.theforeman.org/issues/33369