August 18, 2006

LinuxWorld wraps up

Author: Joe 'Zonker' Brockmeier and Robin 'Roblimo' Miller

SAN FRANCISCO -- Another LinuxWorld Conference & Expo has come and gone. More than 10,000 attendees and 175 exhibitors passed through the Moscone Center in San Francisco. The show ended yesterday after a shorter schedule of talks and exhibits, including a session by kernel developer Greg Kroah-Hartman. Here's our final report, including more video goodness.

Walking the floor of the Moscone Center this year, you might have the impression that the show was a bit bigger. Actually, the show was held in a different building, and the actual floor space was reduced. IDG vice president Melinda Kendall says attendance at the show was flat, so the Linux community held its own in 2006, even without Red Hat to draw in attendees.

Now that the show has come to a close, it seems that the biggest news was what wasn't happening -- that is, Red Hat's absence from the show. Even Thursday, a lot of people were still talking about it, and many attendees felt a bit put out that the company chose not to make an appearance.

Troubleshooting Linux

Thursday morning, Christina Noren, vice president of product management with Splunk, presented a talk on "Troubleshooting Linux and the Open Source Software Stack." Originally, Michael Baum, Splunk's "chief executive Splunker," was scheduled to do the talk, but Baum was unavailable, so Noren stepped in.

The talk contained some useful information and suggestions on troubleshooting open source deployments through copious logging (something any good admin would do anyway), but the talk also contained plenty of plugs for Splunk -- including a plug and demo of Splunk by Corey Shields of the Open Source Lab.

Noren had some good advice for attendees -- plan ahead for troubleshooting by making sure you have the logs you need before there's a problem, use centralized logging so that logs are easy to get to, let users have access to logs if possible to research their own problems, and log success as well as failure so that you've got a baseline to measure against. It would have been nice, though, if the presentation would have been a little less Splunk-heavy, and if had contained some advice for actually solving problems once they were found.

Kernel version control with Quilt

During the last session, Kroah-Hartman gave a presentation on doing kernel version control with Quilt, Ketchup, and Git. As it turned out, Quilt and Git are actually useful for other projects as well, and Ketchup also looks like it could be useful for admins, so the presentation was of value for those of us who aren't kernel developers.

Ketchup is a tool for grabbing kernel source trees directly from the mirror system, without mucking about with tarballs and patches by hand -- it does all of that automagically. This is a tool that would be of great use for admins who build their own kernels, as well as developers who write and test kernel code.

Kroah-Hartman then demonstrated Quilt, a tool for working with patches to make diffs and send emails with those patches to other kernel developers. Though Quilt was created for kernel development, Kroah-Hartman says that it's perfectly suitable for other projects as well.

After Quilt, Kroah-Hartman talked about Git, which he says is "one of the most undersold tools out there." He says that Git is "by far the best" distributed version control system available in terms of speed. Kroah-Hartman discussed the way that kernel developers use Git to maintain their own trees, and showed one of the GUI tools for Git that shows changes to the source code as well as the "path of blame" for patches, so that if something breaks, they'll know who submitted the patch and signed off on it. Again, Git is useful for projects besides the kernel, so any development project could use Git.

Kroah-Hartman had the unenviable "last session of the conference slot," which means that a huge number of attendees were already on their way out of town, or enjoying a few end-of-show drinks with friends and colleagues. About 20 people showed up to hear Kroah-Hartman, and there was a lively Q&A after his talk about kernel development in general.

One attendee wanted to know about the Reiser 4 situation, and Kroah-Hartman says that the Reiser folks "have not read the documentation/HOWTO" on how to submit a patch, who to submit it to, and coding styles. Though it's been a controversial topic, with accusations of playing politics or favoritism for Ext4, Kroah-Hartman says that it's just that the Reiser developers had not been following the well-established rules to submit new code to the kernel. He also says that they are following the procedures now, so perhaps we will see Reiser 4 in the kernel sometime in the nearish future.

The famous "what if Linus were hit by a bus" question came up during the session as well. Kroah-Hartman says there's no specific successor for Linus, but that the kernel development structure is sound enough to continue producing the kernel even if Linus were out of the picture.

Overall, LinuxWorld was a decent show, if a bit lacking in excitement. The quality of the talks was down a bit this year, with a Kroah-Hartman's talk Thursday being one of the obvious exceptions. The next LinuxWorld will be held in New York, February 14-15, 2007.

Click to play video
Click to play video
Click to play video

Previous coverage

You can find our previous coverage this year's LWCE here:

Click Here!