February 27, 2003

Sickness fading

My Bronchitis seems to be fading. It’s not gone yet: I’ve still got some great drainage, hacking cough, and other symptoms I won’t entertain you with. It’s fading, though. I’m real tired tonight, of course, and it’s only 0348. Blah.

Took it very easy for most of the days this week. I don’t mind working from bed so much. Things to have in bed when “working” and sick: laptop, book you’re currently reading, TV and DVD remotes, cell phone and headset, Kleenex, bottle of water (I used 1L Rubbermaid bottle), bag for garbage, cardboard tube to hit kitty with when he’s being a bad little Mr. Kitty, and maybe some snacks. Too bad I don’t feel comfortable peeing in bed, otherwise I might never need to get up.

Today’s remembered lesson: ip link set DEVNAME name NEWNAME. I think I may have even logged that in here at one point. I was thinking about it tonight when I was thinking of the problem people have when they have two cards that are detected by the same driver and can’t control which one gets assigned eth0, and which gets eth1 (for example). You could probably do it in a little dance with the above command. Voila.

I want RH 8.1 to be released. Phoebe is in its third beta. I want to install new desktop box. Bastards.

I’m practically overwhelmed with work. One client wants documentation for stuff and is planning to do a project involving me in the next month, another has two or so projects going, trying to install new client, have another client that is expecting a firewall/mail server from me ASAP (I know, putting them both together seems weird, but low cost), and then there’s all the usual stuff. Like the things that are broken at work. Like the mail server that isn’t backing up because, apparently, even the second tape is bad. Or the drive is bad. I hate these Seagate “Travan” drives. They’re shit. Especially the IDE versions: avoid them like the plague.

ObWebLog being basically complete, I think I’m going to go get ready for bed. Tonight’s “darky is sick so he turns on movies as he goes to sleep” flick is… The Matrix.

… Oh, I almost forgot to talk about that DHCP problem I last logged. As it turns out it looks like dhcpcd will reused a cached lease forever if it’s lease time is (surprise) “infinite.” After trying to get ISC dhcpd to hand out infinite leases by using the decimal equivalent of 0xffffffff seconds, I found out that somewhere in dhcpd 3.0 they stuck in the infinite keyword. So I think I have default-lease-time infinite; max-lease-time infinite; in there now. Haven’t tested this at work yet, though. Need to basically block the DHCP server from a particular server, then see what happens when dhcpcd tries to renew. I’ll do it on a Masterswitch’ed server, actually, and leave it running for a few days to see if it crashes. Really, if it were going to crash I’d expect it to do so within a few hours, like the servers apparently did the night the power went out and dhcpd didn’t come back up.

February 24, 2003

I need to update my anti-virus software

I’m sick. Again. I was just sick a couple of weeks ago, though it only lasted for about two days. Now I’m pretty sure I’ve got Bronchitis. I’m betting that’s what euphorik had/has, and I found out that darkho probably had it to and is just getting over it. I’ve been feeding on Chicken Noodle soup for the past few days. Sigh. I hope this passes soon, I’ve lots of work to get to. When I’m sick I prefer to rest a lot in hopes that it’ll help me get back to complete wellness faster.

I sent out messages to those who requested DD soundtrack notification today. If you didn’t get yours, mail me and I’ll see what I can do.

iptables connmark matcher has problems with iptables-save/iptables-restore, and its command line parsing. Here’s my connmark patch, applied against RH 8.0′s iptables 1.2.6a.

Had a major problem with the network at the office last night. Power apparently went out long enough to bring everything down, including the stuff on UPS. Then when the firewall came back up the DHCP server wouldn’t start up because I never completed the IP switch-over on it. Now lets remember that I switched all the servers, Linux and Windows alike, to be DHCPed a month or two back. Result? dhcpcd and probably pump give up trying to get a lease after a while, and so the servers never got IPs. Oops. I’m going to be changing the time they spend trying to get a lease to probably extend into infinity, but I’m not sure this is entirely good enough. I’d really like them to boot up, see they’ve still got a valid lease, and use that lease – especially in the event they can’t contact a DHCP server. Maybe that’ll be a trivial hack to dhcpcd, who knows? (Of course, for all I know they boot, try and get a lease, and if dhcpcd times out it forks in to the background and keeps trying to get a lease for whatever timeout it’s configured for. The alternative is that it’s going to be booting and sit where it tries to bring up the network interface until it successfully contacts the DHCP server. I guess that wouldn’t be so bad. At least it’d be a start.) Anyway, next I’ll be attacking dhcpcd, or at least reading its source to see how it uses an option or two.

It was fun taking cold medicine, then getting a call from three ISPs (we had a customer line go out too; sounded like TWTC left me a voice mail using a soup can and some string) about 30m later to tell me everything was down. At first I was like, “well, I guess I can drive to the office.” A minute or so later I realized, “I can’t even focus on the monitor, like hell I’m going to drive.” Had to go in today, though, and it was taxing.

Computers are such a bother.

February 19, 2003

Posted tagging standards

So my real big news for the day is that I’ve posted my tagging standards. I think this even calls for an index file on http://www.codefu.org/, which is currently barren. I can see already that there are broken images at the top of the latex2html generated HTML. It would be nice to apply a bit of CSS to the index for that directory too. Oh, and probably critically I’ve forgotten to include some of the TeX files that are necessary to build this. Maybe I should just remove the TeX versions, really. For now, though, it’ll do I think.

I’ll also take this time to note that, in Yellow Dog Linux 2.3, the tetex-latex package appears to be missing latex2html. Bastards.

I suppose others might not find all that as exciting as I do. I did find a Helium2 1.2.1 crack today as well. I’m not using it of course, because that would be illegal. I found it on EFnet, and that’s all I’m going to say about it. I tried using Helium2 for a while tonight (the trial, that is). It’s gotten better than when I last tried it (1.1), and has Ogg Vorbis support supposedly, but still has weird quirks. For example, I had “Copy ID3v1 -> ID3v2″ or something like that turned on (as it is by default) and kept getting my ID3v2 track title copied into my ID3v1 tag. The real problem was that there was no possible way you could fit the entire ID3v2 field in to the ID3v1 field: it was too long. Helium2 did it anyway. I figured out how to stop it from doing that before I was finished with the file, so I didn’t check to see if it would actually write out an invalid tag or not. Other than that, it appears to be quite usable, and have all the features you want. The CDDB wizard even worked pretty well on the two albums I tried it with. Too bad it’s not free, and too bad no one has outdone it with something free.

I found out about a new satellite connection at one of the sites for our new client. The one we were supposed to install with this weekend. I don’t think that date is very likely, since they decided that it needed to be kicked to their “consulting team.” To me this sounds like, “we need a piece of the pie too,” and frankly that’s fine with me. Of course, our client might not be as easy going as me.

Enough for now, must sleep. Before I do that I think I’ll start DD Vorbis files uploading.

February 18, 2003

OpenH323 vs. RH8, standard tagging for MP3 and Vorbis

Doofus is trying to get me to help him test using NetMeeting for a client of ours which is sort of interested. In reality it’s the owner of a client, which earns a dubious place in my work queue, but I’ll attempt to attend to it as best possible. That said, apparently the Open Gatekeeper H.323 Proxy at least doesn’t like to compile under GCC 3.2 apparently, which is what RH8 includes. This project requires the OpenH323 project’s libraries as well as their (I think it’s theirs) pwlib. Both of the versions distributed from the SourceForge.net project I mentioned above don’t really seem interested in compiling in GCC 3.2 either. RH8 does include packages named openh323, openh323-devel, and pwlib. I tried installing those and then compiling the H.323 proxy mentioned above, when I learned that the proxy had basically the same errors that the older versions of the OpenH323 stuff included. I think the OpenH323 project is supposed to include some binaries too, but they’re not in the openh323 package that I’m getting in RH8, so maybe they’re just not present in RH8. Or I hallucinated them. Final note: there is a pwlib-devel package that I didn’t install, but should have. Grab all four if you want to try to do anything even remotely useful, I suspect.

I’ve downloaded several albums over the past few days. Albums of note currently: 11 Beatles albums (all the major ones, excepting one or two I suppose, such as Yellow Submarine which I suspect I hate), The Apples In Stereo Velocity of Sound, Cocteau Twins Treasure, Cowboy Junkies The Trinity Session, and Liz Phair Whip-smart. This is not all I’ve downloaded, just the ones I like the most right now. (I just put on Magical Mystery Tour.) Looks like a total of 28 albums so far, mostly off Soulseek which I find to be quite nice. I wish it had the ability to filter based on bit rate, though.

Given this wealth of new albums, not even counting those that have been sitting on my drive untended for many months, I opted to go back and make some modifications to some documents I made that (basically) standardize ID3 tags. I’ll be posting them up here soon, maybe tonight if I finish them.

I really need to post the DD soundtrack too. I know. :(

February 12, 2003

The Cantenna lives!

I went about fixing my Cantenna today. I made the hole much bigger than the element, and re-soldered the element to the N connector. Results are quite good. So good, in fact, it justified me going out and buying a tripod to mount it on. Per the pictures from the Cantenna page I just secured it with a random piece of metal I had (one of the pieces you take out when breaking in to a new 5.25″ bay on a PC case) lying around and some zip ties. From the back of my apartment I can see the AP in the back of ardent’s apartment. Just the results I was looking for. Long live the Cantenna.

In other news, and in light of my recent exploits with the BIOS, I decided to get ballsy and edit the BIOS on my laptop. I wanted to keep it from suspending when I closed the lid. There was no option in the BIOS by default, and there was no way to easily disable the switch short of breaking it off. (At least, none that I could see.) BTW, the laptop I was working on was a Twinhead P98 laptop. I’m not sure which variety it was, but I suspect it to be a Slimnote GX3 or some such. (While I’m rattling off model numbers, I believe it has a MX29F002T flash chip. This according to AMIFLASH.COM; see below.) I figure there’s probably either an option in the BIOS that I can enable to keep it from suspending, or maybe I can actually replace the “Power Management” section of the BIOS with that from another similar laptop that does have the functionality. (That second one actually makes me laugh in retrospect. What are the chances of that working?)

For starters, none of the three or so AMI flash utilities that I had been using was able to auto-detect the chipset and part model numbers for the laptop. I also tried the latest UniFlash. Or, at least, I thought this was the latest at the time; now I see that there is apparently a newer UniFlash which I did not try. Anyway, the UniFlash I tried (1.17b I think?) gave me a runtime error and was generally useless. So I downloaded the P98TF 1.07 AMI BIOS from Twinhead. This BIOS came with a program called AMIFLASH.COM which is a very small file that appears to be good for precisely one thing: flashing the BIOS. I suspect it can backup too but I didn’t bother checking in to what incantation invokes this behavior.

I jammed the BIOS image, AMIFLASH.COM, and my trusty AMIBCP75.EXE onto a floppy disk. Booted in to DOS. Used AMIBCP75.EXE on the BIOS image and enabled a whole bunch of options — including one to apparently turn off APM all together. Flashed it onto the laptop with AMIFLASH.COM and rebooted. (There was actually some praying in there, but I digress.) Wonderfully my modifications had worked; the menu items I had enabled were all there. Some of them seemed kind of inaccessible, as in grayed out, but the important ones were there. I disabled ACPI which then allowed me to disable APM. There was no option for “don’t suspend when I close the lid.”

I’ll take this time to note that these notebooks are (A) cheap as shit – and you’ll get what you pay for; and (B) these notebooks tend to eat fucking batteries. Something inside goes coo-coo for Cocoa Puffs and apparently destroys the charging circuitry. This in turn destroys your battery within a few hours, I’m told. I’ve never observed this because I manage to only get the laptops after this has happened and they’ve already got absolutely no chance of battery power. In other news one of the hinges for the screen completely broke off tonight, threatening to make the laptop just that much more useless.

I will say that after disabling these options, I was able to boot into Linux and keep it running with the lid closed. No problem.

Installation tomorrow morning. (OK, OK, so it’s really at 1200. That is my morning.) I guess I’ll head over to bed.

Oh, I almost forgot. I talked to the person that I was ranting about last night, the one who was supposed to flash BIOSes. He basically told me that no, he didn’t flash the computers, and no, he didn’t say he had. He said I needed to test it and tell him if it was OK, then he’d go flash everything. I was kind of perturbed since I think he was going to flash them then give me a flashed one to look at, but then found himself unable to flash it. So rather than tell me, “hey, I can’t make this work,” he changes to, “hey, go ahead and test your BIOS.” It bothers me when I perceive that people aren’t admitting fault, perhaps because I’m so ready and willing to do it myself. Anyway, another one slips through my fingers.

Uh. Hm. I just realized that I have no idea how to get to the new customer I’m going to tomorrow.