KernelTrap has a post up about the clustering file system that Matthew Dillon is designing.  That’s not news to regular readers here, but there was an interesting comment on the story about an existing clustered filesystem called “Lustre“.

iXSystems has bought FreeBSDMall.  Interesting, as now it is a company with its own software and hardware product.  There’s only a few others like that, who coincidentally also have offered BSD-based products.  (Seen on BSDNews.)

Matthew Dillon’s written up a description of how indexing (via B+trees) will be handled in his as-of-yet-unnamed clustering file system.

AsiaBSDCon 2007 has no DragonFly-specific events happening at this point – Masao Uebayashi posted a neighborly welcome for any DragonFly developers to join the NetBSD discussion and beer party on March 9th, starting at 13:00.

OSNews has a news story up about Matthew Dillon’s planned new filesystem, with comments of varying utility.

I didn’t know this, but it’s very simple to assign VLANs to your network interfaces in DragonFly.

YONETANI Tomokazu happens to know the secret to building multiple kernels at once.

This week on UnixReview.com: the Regular Expressions article “Tcl Scores High in RE Performance“, “Examining the Novell Certified Linux Engineer 10 Certification“, “Test Your Knowledge of HTML Topics” in Q&A format, a book review of “The Relational Database Dictionary“, and a very silly “Codysseus: A Geek Travesty by Erudil“.

Sascha Wildner has a version of DragonFly 1.8 compiled using NATA, the new ATA system, available on leaf.dragonflybsd.org.  (Link goes directly to a bzipped ISO)  Try it if you’ve had trouble getting DragonFly to install on a system with a very new SATA controller.

“I’m BSD” – linked by Thomas Spanjaard.

Matthew Dillon has written an extensive description of the plan for the as-yet-unnamed Dragonfly clustering file system.

Francois Tigeot managed to get pkgsrc’s wip/jdk14 built on DragonFly. That’s great news, as the JDK is in demand, but one caveat: he did it on version 1.4 of DragonFly. It looks like some change, perhaps in system compiler, is the cause.

Update: How to duplicate his efforts.

Matthew Dillon went into more details about what he plans for the as yet unwritten DragonFly file system.  (dfs?)

It looks like Matthew Dillon is going for his own filesystem to meet DragonFly’s clustering goals.  According to his recent post, it will be a (deep breath) 64-bit fsckless logged filesystem with collapsible snapshots limited only by disk space, local caching, and logical-level multi-master replication.  It should have a lasting effect. ETA: summer release, in some form.

Thomas E. Spanjaard has NATA, the ‘new ATA’ code from FreeBSD, in the DragonFly source tree.  It’s in, but not yet enabled, in release 1.8.  If you want to try it out, or if you have certain newer hardware that demands it, he’s written how a document on how to enable NATA, on the wiki.

An ongoing conversation about ZFS and whatever distributed filesystem Matthew Dillon potentially writes led to some interesting links from Bill Hacker and Rupert Pigott: IBM’s GPFS, GridAFS, and Distributed Shared Memory.

Seen several places: an article about using NULLFS, the BSD filesystem that enables the mounting of a filesystem over itself.  If that sounds confusing, perhaps you should read the article.  Conveniently, NULLFS in DragonFly was recently re-enabled.

OnLAMP/BSD has a new article up comparing a few firewalls.  It only mentions “OpenBSD” as a software firewall, though what it’s really talking about is PF, which DragonFly also uses.

It wouldn’t be hard.  There’s only two steps, neither of which are new, and it would be cool in a super-nerdy way to be able to save the state of an entire running virtual kernel.

If you like setting various compiler options at some risk to stability, Alexander Shiryaev has found a list of suggestions. Not surprisingly, it’s a Gentoo Linux page, and there’s also some caveats.

Seen a number of places: I’m BSD.

From Brett Estrade on IRC: the UC Berkeley EECS school has a recent paper out that talks about massively parallel systems and the future needs of those platforms. (That’s where DragonFly is going – parallel – and where it’s been – Berkeley.)

Geoff Speicher wrote a very nice transcript of the recent BSDTalk interview featuring Matthew Dillon.

Matthew Dillon has gone into some detail on his thoughts on what a clustering filesystem needs.

Via waxy.org, a textfiles blog entry that describes the structure and events that happen to online communities.  Some of it can be thought to apply to DragonFly or other BSDs.

Preview, the version of DragonFly inbetween released code and bleeding edge, has been slipped forward.

If you’re following the released version of DragonFly, the ‘slip’ tag (check your cvsup config file) is the most conservative possible way to update.  Speaking of which, a 1.8.1 release is due soon.

UnixReview has several new articles up: “Sprints Important in Open-Source World“,
a certification article: “Examining the Novell Certified Linux Engineer 10 Certification“, and
a book review of “WarDriving and Wireless Pen Testing“

BSDTalk has an interview of Matthew Dillon up, talking about the recent 1.8 release. (First spotted by Sepherosa Ziehau)

If you have a desire to help out non-native English speakers, a transcription would be useful.

AsiaBSDCon 2007 is March 8th through 11th in Tokyo this year, and the schedule is now available.

It’s very easy to debug a virtual kernel, using gdb, with one caveat/fix.

‘timofonic’ linked to Luigi Rizzo’s work on emulating Linux for the purpose of running Linux-specific device drivers, kind of like Project Evil.

It’s not possible yet, but Matthew Dillon outlined the steps needed to get checkpointing and virtual kernels working together – you could start a kernel, and ‘freeze’ its state – even sending the resultant file to someone to restart and debug.

Matthew Dillon mentioned that he was trying to decide on what filesystem to use that would help with clustering.  The general consensus is “ZFS“, but other filesystems entered the conversation, like LFS (still working on NetBSD) and Plan 9.  Also: an explanation of filesystem snapshots.