It’s possible to speed up a ‘make buildworld’ by increasing the number of parallel make processes, with the -j option. However, the optimal number of make processes depends on your system setup. Simon ‘corecode’ Schubert did some testing, and it looks like the number of CPUs +1 is the best option – as long as you have more than 1 CPU. His writeup even includes a nice graph.
Johannes Hofmann has taken over estd, a “frequency scaling daemon for NetBSD and DragonFly”. The newest release brings multicore support on DragonFly.
This description of a Hammer bug makes for interesting reading, since it delves into the sequence of events where data is actually laid down on disk. Interesting reading for a geek, admittedly…
Do you have an Acer Aspire One D150 and you want DragonFly to work on it? Alexander Polakov has one something similar, and he got it to boot. Sound and wireless networking works too.
Nuno Antunes has a compact writeup of what it would take to finish porting netgraph7 to DragonFly, if you’re interested…
The build for pkgsrc-2009Q3 packages performed on i386 DragonFly 2.5 is complete. There’s a build log. These packages are immediately available if you are on a 2.5/i386 system and use pkg_radd. If you want to upgrade, try pkgin or pkg_rolling-replace.
Version 3 of Hammer is now available in bleeding-edge DragonFly, though it’s still experimental. The biggest reason for this version bump is to move the /snapshots folder to /var for all Hammer filesystems. This means an accidental <tt>rm -rf</tt> won’t destroy snapshots, as I’ve done. The saved data is still on the original partition, as just the metadata is saved to /var. More explication is available.
Jan Lentfer performed some Postgres benchmarks on DragonFly. It’s elaborate enough that it’s in the form of a PDF attached to the message I’ve linked. There’s some additional variations that haven’t been tried yet.
Vigorous file system activity seemed to lower performance in the long term on Hammer, which is certainly something to investigate. More testing please!
Stathis Kamperis has written up a description of the test framework he designed during the Summer of Code. It may end up in DragonFly, which seems like a good idea to me. It’s designed to be generally operating-system independent. He includes a link to the git repo where he’s keeping it now.
Can you think of something that:
- Takes about 4-6 months to do?
- Can be used in DragonFly?
- Is usable as a Computer Science thesis?
Francois Tigeot reports having used vkernels in production quite successfully to isolate some legacy software, even though vkernels were only planned as a development tool. Nice to hear of something being more useful than intended.
There are now official but experimental git repositories of pkgsrc available. One’s already available for DragonFly, but either should work.
Details of the new release are found in the announcement, including some biggies like KDE4. I’m building binaries for this release, for DragonFly i386/2.4, i386/2.5, and amd64/2.5. (Though the 2.5 binaries for amd64 should work on the amd64 2.4 release, too.)
Luiz Gustavo made a screencast for the DragonFly installation process. It’s 15 minutes long, requires Flash, and it would probably help if you speak Portuguese. I like seeing videos like these; multiple media always make for more fun.
Stathis Kamperis has ported POSIX message queues to DragonFly (from NetBSD) and has his eyes set on veriexec next.
The release announcement isn’t out, but the branch is there. I’m building it for DragonFly 2.4 and DragonFly 2.5 on i386 now, so we should have binary packages in about a week. I should have reports to go with it.
If you use NFS, especially with vkernels, you may be interested in the latest round of NFS changes recently committed.
The next quarterly release of pkgsrc should be released by next week. Normally it is released 2 weeks after starting a freeze period, but this release was slightly delayed for some structural changes and for KDE4.
Matthew Dillon solved a performance problem that was most noticeable when doing intensive I/O while performing other tasks; downloading a large collection of files while opening another application that read a lot of initial data, for example, would have a noticeable startup delay. His recent VM change seems to have solved it, and the commit message has an in-depth explanation of how.
Sdävtaker has reported success booting DragonFly on a Clevo TN120R, which is almost more of a tablet/nettop than a laptop.