SambaXP 2017: John Hixson’s Reflection

 

Samba XP 2017 took place in Gottingen, Germany. I arrived on Monday evening and was greeted by Ash and Kevin Kunkel from Indeed.com. We talked about Open Source and Active Directory over several beers. Kevin maintains what is probably the largest Samba AD in the world.

On Tuesday, I attended a tutorial on how to create a Samba4 AD with DHCP and Dynamic DNS. While I have done this many times, it was nice to see this from someone else’s perspective. Much was learned!

Wednesday was the first day of the conference. The conference opened up to Jeremy Allison thanking everyone for coming and a little bit of talk about Samba.

The first talk was given by Tom Talpey of Microsoft. It was about future directions between Samba and Microsoft. It was an excellent talk. Tom went over things like clustering and distributed filesystems as well as how Microsoft interacts with the Samba team.

The next talk was given by Jeremy Allison on the recent symlink CVE. Jeremy explained how it was discovered and the measures that were taken to fix it.

Kal Blin gave several examples of a Kerberos transaction in various programming languages to demonstrate what languages might best be used for Samba in the future. The concern here is the the current and next generation of programmers are not learning C, so we must focus on a language that they do know.

Petr Viktorin spoke about the ongoing migration of python2 Samba code to Python3.

Ralph Bohme and Douglas Bagnall both spoke about Samba performance, measuring it, tools to use for doing so, and improvements that could be made. The bulk of these talks used Linux tools and practices. I really think we need to get some BSD into this conference 😉 .

Andrew Bartlett spoke about scaling Samba AD domains to 100k users. This was another talk that went over areas of Samba that need improvement or have improved and the methods that were used to get there.

After Andrew’s talk, there was a social event in in downtown Gottingen near the city center. I had several good discussions with several people involved in the Samba community. I met Chris Hertel who wrote the book “Implementing CIFS” and talked with him about the SMB protocol. I spoke with Andrew Bartlett about improving Samba on FreeBSD as well as fixing many outstanding issues Samba has on FreeBSD. I met the two crazy French guys from Tranquil IT systems – Vincent & Denis Cardon. They were a blast and told me about their company and the various environments they support.

Day 2 of the conference started a little hungover, but I was a trooper 😉 .

The first talk was given by Alexander Bokovoy from Red Hat and was about implementing a global catalog in FreeIPA. I find it very interesting, but not much use for FreeNAS (but possibly other products should we ever implement them 😉 ) .

The following talk was about the Samba team switching to MIT Kerberos. Apparently this switch allows enterprises to ship distributions. I do recall Heimdal Kerberos had various restrictions due to its crypto.

Stefan Metzmacher from Sernet gave a talk giving more details than you ever wanted to now about how Windows authentication works. Even I could barely stay awake through this, yet, it was very detailed 😉 .

Jeremy Allison gave a talk on how to write a samba VFS module. This was by far my favorite talk during the conference. I’ve written several now, but getting up and running was non-trivial since there is little to no documentation on doing this. Jeremey went over everything in fantastic detail and answered lots of good questions to make it clear.

Kevin Kunkel from Indeed.com gave a talk on his large Samba AD environment and the various problems they have. He described the growing pains of doing so and addressed the Samba team directly with various questions and issues that he would like them to resolve.

Denis Cardon described pretty much everything I spoke to him the night before in his talk.

The last talk was more of a panel discussion and was about SMB3 and clustering. The bulk of this disucssion was about persistent file handles. We could learn a lot from this should we figure out a clustering strategy.

 

 

 

 

John Hixson, Senior Software Engineer

Submit a Comment

Your email address will not be published. Required fields are marked *