On Wednesday the 4th of march 2015, Logilab hosted a sprint on salt on the same day as the sprint at SaltConf15. 7 people joined in and hacked on salt for a few hours. We collaboratively chose some subjects on a pad which is still available.
We started off by familiarising those who had never used them to using tests in salt. Some of us tried to run the tests via tox which didn't work any more, a fix was found and will be submitted to the project.
We organised in teams.
Boris & Julien looked at the authorisation code and wrote a few issues (minion enumeration, acl documentation). On saltpad (client side) they modified the targeting to adapt to the permissions that the salt-api sends back.
We discussed the salt permission model (external_auth) : where should the filter happen ? the master ? should the minion receive information about authorisation and not execute what is being asked for ? Boris will summarise some of the discussion about authorisations in a new issue.
Sofian worked on some unification on execution modules (refresh_db which will be ignored for the modules that don't understand that). He will submit a pull request in the next few days.
Georges & Paul added some tests to hg_pillar, the test creates a mercurial repository, adds a top.sls and a file and checks that they are visible. Here is the diff. They had some problems while debugging the tests.
David & Arthur implemented the execution module for managing postgresql clusters (create, list, exists, remove) in debian. A pull request was submitted by the end of the day. A state module should follow shortly. On the way we removed some dead code in the postgres module.
All in all, we had some interesting discussions about salt, it's architecture, shared tips about developing and using it and managed to get some code done. Thanks to all for participating and hopefully we'll sprint again soon...