It's almost at the same period of the year when it happens and at the same location (almost). It's the third time in a row for me and still I have the same pleasure to get there work hard on Samba, exchange ideas and anecdotes with other team members and unfortunately not sleep not to much.
What makes this event quite unique and productive is that we usually have a limited amount of goals that we want to achieve this week and the presence of ~8 team members surrounded by Microsoft's engineers create this emulation (well at least that's my theory !).
Last years during this event we :
- added support for windows joinning a Samba 4 AD domain
- added support for RODC
- added partial support for multi-domain forest
This year no big title: "Samba 4 now support xxx", not because we weren't able to achieve some results by then end of this week, but just because since 9 months we are focusing on releasing the first stable 4.0 release.
So we decided to this opportunity to torture Samba against series of test suites written by Microsoft so that we can be more confident that we will effectively deliver a stable version which works for most of our users.
The good (or should I say the great as we are in US ?)
Without going to deeply in the protocols (or more exactly in the document naming related to those protocols), we did very well in most of the tests with very small problems on most protocols we tested.
That's a good news because not only Samba pass our own torture tests, it also pass the tests that are based on the protocols documentation, and on some protocols reading them is already a torture so I let you imagine testing them !
We found also lots of bugs (check the ~25 bugs in a row) and managed to increase by 50% the number of successful tests on the test suite (related to the Directory Replication Service) protocol that was badly failing.
The bad (but not so much)
So DRS is the test where we had most of the failures, some are expected as we don't have the features, some were just related to wrong error code and some weren't expected. Still after this week we have a better view of what is not ok and given our current focus that's pretty reassuring.
Samba 4.0 focus on the single domain forest, as we don't have support for sysvol replication multi-dc with mixed Samba and Windows DC is not supported and even pure Samba DCs leave administrator with the challenge on how to solve this replication problem. Tests results seems to indicate that nothing critical prevents the latter scenario (once you figured out how to replicate the sysvol share).
That's also good news because even if not officially supported we know that some people will try it and know that it won't blow in people's hand is quite relaxing.
Well there isn't !It not perfect this week was a success till the point that we had sunshine every day in Redmond (and no rain).
What's next ?
Obviously we have a couple of stuff to fix before we can declare a stable 4.0, then we will all take some rest and see what we will do next year at the same period of the year ...