What should Ansilog do? (Product requirements)

[yakirh]

• What are the features? What features can wait for the next version?
• Can users login with github user?
• Will there be registration?
• Will there be option to STOP build?
• Option to support builds without MINOR and MAJOR? (e.g. in Ongage the Staging Next build doesn't have MINOR and MAJOR, just incrememnted number of builds since the dawn of the project.

[rafi]

In my honest opinion:

• Minimum requirements: Run multiple playbooks on multiple hosts in a central centralized location with the ability to view output of previously executed playbook jobs. Issue Entities #9 discusses enterprise rules.
• Authentication: Several options: basic-auth (browser dialog), users db table, social login (github, facebook, etc). Issue Authentication #5 discusses authentication options, and Authorization / Permissions #6 discusses authorization.
• Build/play control: Start, start with options, stop. Issue Concurrency handling and locking #3 discusses several business rules.
• Versioning: MAJOR.MINOR.PATCH is a semantic convention called semver. A project must have a versioning scheme, the only hard-coded integer a build system can provide by itself is the job/build serial number.

[rafi]

Run multiple playbooks on multiple hosts in a central centralized location with the ability to view output of previously executed playbook jobs.

• Quickly listing jobs and jobs' builds.
• Allow viewing previously executed playbooks' console output.
• Clear notice of build status: success or failure.
• Deploy to machine by version # using auto-complete from Ansible hosts and git tags.

[adamli]

• Gather and Highlight failed tasks, in hipchat and in related job

[ronenli]

• Import playbooks from file / git repository (including playbook variables, roles and tasks)
• In case playbooks are imported from git repo, keep them syncronized. (?) Should it happen using an ad hock action OR automatically when updating the repo)
• Allow using ansilog project's playbooks (provision)

[rafi]

What about history, i.e. jobs that were executed (job+execute=build?) multiple times, and data retention?
These are questions we should ask:

• Do we need more than one last log for each playbook?
• What does 'executing/running a playbook' mean? what happens besides actually running the playbook?