The Treehouse Blog

In preparation of Doug‘s upcoming LAN party I’ve had to order some parts for computer upgrades. I’m currently running a GeForce 2 MX 400 (64 MB) which is a few generations behind the times, and it started showing its discontent for UT2k3 at last year’s LAN party. Since there is a UT2k4 out this year, I’m clearly doomed. So, I ordered a GeForce 4 TI 4200 (128 MB), a step back from Ian‘s 4600, but should still prove acceptable. Also… realizing that UT2k4 needs something like 5 GB of drive space, I had to do something about the 30 GB drive in Pine. I considered augmenting it with the 10 GB in my FreeBSD/Solaris x86 test box, but that seemed pretty horrible. Trying to run the game off of the network looked equally grim. So… back to pricewatch it was, and a new 80GB SATA drive should be headed my way. Pine should be happy.

Of course, Pine isn’t the only thing complaining. Oak is a bit underpowered when it comes to things like, say, transferring files at 200Mbits/sec, or compiling NetMRG. So, much to Pine’s chagrin, parts were ordered to make Oak almost as powerful. I am pretty ticked that I had to pay more for RAM than last year… and it’s even not as good of RAM. So, 7 independent orders later (worst case of order fragmentation I’ve ever had) things should be fine…

Well, sorry to bore you with that, but I spent far too much time on this today to not devote an entire blog to it 😉

The gigE card arrived today. The kernel shipping with FC2 has broken drivers for it, but the updates take care of it. I was able to push about 15 MB/s out it to my switch (same file to two machines) which pegged the CPU on Oak. Upgrade in the works, but not sure to what.

I’ve pretty much given up on S3 ACPI suspend on my Thinkpad R40, despite getting a kernel compiled that mostly works after resume, this suspend mode seems to use way too much battery. I’ve got the swsusp stuff working tho, slow but effective.

Done with DS9, too.

So, instead of watching DS9 or making my laptop more usable, I spent last evening trying to do APRS in Linux. Linux has drivers for using your soundcard for doing AFSK1200, but they are hardware-dependent, requiring a plain SoundBlaster or Windows Sound System-compatible card. Since I don’t believe I have any of these anywhere at the moment, let alone in a usable or mobile form, I started looking for alternatives. Soundmodem is a user-mode sound card interface for Linux that can do various encodings and interfaces with the Linux AX.25 stack, or can behave as a TNC sitting on a pty. It seems to work really well (and has awesome configuration tools) but it also consumes 25% CPU on my 1.5GHz Pentium-M… which is way too much, I think. Multimon, which is written by the same guy, uses almost no CPU, but can’t do transmission or interfacing with other things, just provides text output of what it decodes. The oscope feature it has crashes my box badly if you try to close it… no clue why, but I’m teaching myself to always disable that option.

Doug and I were talking on Thursday about the slowness of the Linux boot process, and it seems some guy at IBM has already discribed the parallel service starting that we were theorizing. I’ll have to ask Ian if Gentoo has implemented anything like this yet.

I won a Netgear GA620 (1000BaseSX) card on eBay last night for <$10. If works out OK, I might need to get one or two more. It seems appropriate that I’ll have gigabit in my house before CTI does at Trindle Commons… Fiber cable looks pretty cheap on eBay too, but there’s more singlemode than multimode it seems.

…And Ship’s 24TB 1.2M$ SAN has been moved from the basement of a dining hall to the computer center. Monday we’re moving all of KLN’s Sun boxes in preparation of its installation.

This has been a really busy week, and it’s only Wednesday. The rest of the week doesn’t hold much of chance of calming down either.

The laptop is in! The Thinkpad R40 arrived Friday (after ordering on Wednesday night, pretty good… about worth the $45 in shipping). I visited both Ian and Ship to acquire media and networking resources required to bring it up to speed. It still hasn’t replaced spruce (no files have moved), so it probably won’t be seen outside of my house much until it does. First three thoughts about the laptop: fast, big, quiet. The first and third are obviously completely good… the second is a mixed blessing. Complete article to come at some point….

Other stuff is happening too. We had a family picnic on Saturday. VBS is this week (Sunday-Thursday) and I’m running the sound/lights for the program as I’ve done for a number of years now. Friday night Isaac and his SO are planning a CTF get-together. Saturday and Sunday are ARRL Field Day. (Need to drag Doug to the last two.) Sunday is the conculding VBS program, as well as a more major family reunion at Doubling Gap, during which Chris and I can hopefully uncover a large quantity of geocaches.

So, not much NetMRG is getting done. My uncle who works for the navy base said that he had come across NetMRG while doing some RRDTOOL searching… might order a T-Shirt ;). This week is work, VBS, and a few hours of DS9 watched on the new laptop to gauge battery performance (as a primary objective, anyway). Some interpersonal stuff, too. Evil ambiguity. And… Blazer doesn’t seem well this week, and Katelyn’s at field hockey camp, and Ship is getting a 24TB SAN, and… well, you get the picture. Busy.

Yeah… so that’s why there isn’t a lannic photo/graph archive yet, nor have I logged the last geocache find…

Chris and I hiked up to a grave on top of a Perry County mountain yesterday in search of a geocache, which we did manage to find. We walked part of a trail in the Hemlocks Natural Area, a very beautiful place indeed. We then trekked up the side of a mountain, walked along its top, and then, about half a mile from our target, we realized that it was on the next mountain over. We had to go through the very narrow valley between, then met up with a trail on the other side. Chris says one mountain per hike is now a requirement.

I’ve been playing with ISC DHCPD a good bit recently. Today I became more aware of the limitations in OMAPI, especially the lack of useful parameters that can be specified for host objects in this manner. This was completely offset by the coolness of dhcp-eval, however.