The May 2010 issue of BSD Magazine is out, with, among other articles, a writeup by yours truly about using HAMMER to access historical data.
I didn’t know about this, but Michael W. Lucas has a new book on the way: Network Flow Analysis. It should be good; his other (BSD-themed, generally) books are surprisingly accessible despite being very technical. (via)
A note, in part for my own benefit: the @reboot crontab entry is all you need to get a HAMMER mirror-stream going again after a reboot/shutdown.
Because of a number of problems, snapshot building hasn’t worked for some days. To fix this, some updates need to happen within DragonFly. This will mean a minor version bump to 2.6.3 in the next little while.
From a commenter on a previous post: Gentoo has a Google Summer of Code project porting portage to DragonFly, by student Naohiro Aota. I had no idea this was happening – this is interesting!
We’ve got 3 projects for Google Summer of Code 2010:
- “Device Mapper based Logical Volume Management”, by Alexander Hornung and mentored by Chuck Tuffli.
- “Porting kernel mode-setting, GEM and KMS, to DragonFlyBSD” by David Shao, mentored by Matthew Dillon
- “Coalesce + MPSAFE kevent, select, poll and wakeup” by Samuel Greear, mentored by Joe Talbott
We had a good number of excellent proposals, but only 3 slots from Google. There were only 12 spare slots by the end of the proposal period, too, meaning less than 1 spare per 10 organizations. I’d encourage people that applied and didn’t get in to still try the work; there were some neat proposals!
Visit the GSoC site for more details.
I’ve made reference to DESTDIR for pkgsrc several times, with only an informal understanding of what it means. From what I’ve learned, and what Joerg Sonnenberger’s told me, DESTDIR support means that packages can be built from pkgsrc without needing to be root. This means local packages can be built on an ordinary user account using pkgsrc.
This also means that pkgsrc can build packages before each upgrade, and only upgrade if a binary package can be built for each item involved. This means minimal downtime and no failures during upgrades, the biggest bugaboo for using pkgsrc that I’ve encountered.
Matthew Dillon went into detail on just how Hammer snapshots could be shared out via Samba.
Two items, via Dru Lavigne: Thunderflash is a new site with images for use in virtual machines, and there could be more of a BSD representation there. Also, if you live near South Carolina in the U.S., Dru could use 4 volunteers at the BSD booth at the SouthEast LinuxFest.
There’s a deficit of vowels in the title of this post… Anyway, Alex Hornung has changed the format of some of the sysctls used in the I/O scheduler, dsched. Be ready for this if you’ve been messing with it in 2.7.
Siju George is making a Hammer volume’s snapshots available through Samba, with the results that some Windows-using developers get historical snapshots for free.
Aggelos Economopoulos has posted his patch for Page Attribute Table support; it needs to be able to perceive CPUs, which is apparently not that problematic. A potential task for others?
Linux Weekly News describes the 2.6 release of DragonFly in an extensive article, which even mentions this Digest. (yay!) Also in the comments, a link to a short interview with Matthew Dillon on a French site with an English translation.
Also, seen on the howling void: Software SSD Cache Implementation for Linux? asks for what we’ve already got: swapcache. Neener neener.
I posted a note about where the 2010Q1 builds are for pkgsrc; if you’re on i386/2.6 right now, you can try it out.
… is one of the better commit message descriptions I’ve heard.
The newest branch of pkgsrc for 2010 is officially out – read the release announcement for details on what’s updated. Among other things, DESTDIR support is almost complete, and a shift to default KDE4 is underway.
I’m working on bulk builds already, so hopefully soon you’ll be able to pkg_radd 2010Q1 packages…
Newegg is running a special on 80G Intel X-25M SSDs – $215, just today. Internet Consensus seems to measure that model of SSD against all others, so it’s a good buy. (assuming you’re living someplace Newegg ships to.) Buy it for swapcache, buy it because it’s a SSD.
Venkatesh Srinivas has been working on new version of DragonFly’s malloc; he’s published an extensive writeup (which is inexplicably split in two in the mail archives) that includes several of my favorite thing: graphs! For those short on attention: the new malloc has around a 20-25% improvement over the existing malloc in MySQL sysbench results.
Matthew Dillon identified a possible data corruption bug in Hammer with a nearly-full filesystem. It’s dramatic enough he’s tagged 2.6.2 and 2.7.2 so that people can update; his message about it describes how to check for corruption.
And the torrent of new activity continues: Alexander Polakov has imported FreeBSD’s mptutil(8), good for managing LSI Fusion-MPT controllers.