I would advise not trying to this with remote_file like that, even if it accidentally works. That's a hack that only addresses file permissions and a use case that we wouldn't want to sign up to support. We could add a "whyrun true" or "audit true" to resources that would cause them to run in "whyrun-always mode" and not do anything even during normal Chef runs and then just output what they would have done, and then resource reporting could report them normally and just flag them as being from an audit resource that didn't do anything but just was reporting. We're aware that Puppet has this functionality but don't really have any visibility into how much that gets used in the Puppet community and if it was considered a hack or something that was useful. The problem with this approach is that its hard to express things like "should not be listening on the telnet port" as a Chef resource so we couldn't cover those kinds of use cases this way. The serverspec language has the ability to express things like this, so there's another idea which is allow users to write serverspec code which would execute as part of recipe execution which would be used to generate these kinds of reports. And then there's two implementation details of this where in one you embed the serverspec code as resources in the Chef recipes and the other where its other files which are added to the cookbook. This would be a use of serverspec in addition to specs used to ci test cookbooks because it would be testing real chef client converges and a lot of the cookbook spec testing code does configuration to exercise edge cases that would be inappropriate to do on a prod server. Daniel, as a user migrating away from Puppet security audits, what would you rather see (and why)? And anyone else who has thoughts... On 9/25/14, 12:09 PM, DV wrote: " type="cite"> |
Archive powered by MHonArc 2.6.16.