Archive for the Uncategorized Category
Sepherosa Ziehau posted a number of benchmarks to show the general improvements in packets-per-second volume with his recent networking changes. As is appropriate when comparing values, he created and links to lots of graphs to illustrate the improvements.
Dru Lavigne has news of the OSS Census now supporting BSD systems for counting; it’s good to participate and provide evidence of the number of BSD users out there.
She also has one of her infrequent but useful link posts up; check it and find more to read.
Martin Schütte is updating syslog (there’s an IETF standard for it, which I did not know) for one of NetBSD’s Google Summer of Code projects; he very kindly posted links to it on the DragonFly users@ list in the hopes that it could benefit DragonFly’s syslogd too.
The Preview tag has been moved up; if you run Preview or 1.12.2, and still have errors building m4 from pkgsrc 2008Q1, add this patch.
Robert Luciani, one of the Summer of Code students for DragonFly, did some initial testing of the libc_r and libthread_xu libraries, with some graphable results. Unfortunately, there’s some degree of error, but that’s OK - I just like having tests performed and images created.
Matthew Dillon’s recent parallelization of cpdup brought up some interesting features: it can do third-party transfers, copying data from one remote machine to another, and while not faster than rsync, it’s relatively easy to use. Vincent Stemen followed up with a mention of his ‘rbu’ (Remote Back Up) product, that serves as a wrapper around rsync and simplifies the backup process.
Sepherosa Ziehau has posted some work he’s done to reduce serializer contention in an effort to improve network forwarding throughput. His detailed technical explanation also includes some benchmarks; he found a way to improve speeds but finds that there’s still a penalty from multiprocessing support.
The USENIX Annual technical Conference is happening June 22-27 in Boston, Massachusetts, and registration for it has opened. Theres a whole lot of events happening, including a separate poster session, so read the link for details. (via)
Matthew Dillon asks, “How can pf be used to create a fair-queue algorithm similar to Cisco’s?“ Answer if you know it; there’s been a few guesses.