Why not? Because all four of those installations were GUI-based, YAST2 installations, using the "Minimal" package selection. According to the manual, the quick-start guide, and the help text on the package selection screen, that option does not install any X components. There is no KDE, no Gnome, not so much as a WindowMaker. And yet, mysteriously, when I boot the system using a Debian rescue floppy and mount the partitions, I find X all over the place. But because the minimal installation doesn't configure X - it doesn't install it, remember? - when it tries to load X, it gets hosed. Why it doesn't then drop out to a command line with a few complaints about not being able to load X, I dunno - and I don't much care. I didn't ask for X in the first place.

I didn't have this problem with the other two servers because they don't have CD-ROM drives. Instead, I installed SuSE on those systems over the network, mounting THOTH's (my laptop) DVD drive via NFS. Doing so requires booting from the installation floppy, loading network drivers, configuring the network, and so on - and by then, you're stuck in YAST1. Which uses a completely different (and in my opinion, much better) package selection strategy. There, if you select a minimal or DMZ-based system, that's what you get. An Xserver is loaded, a very simple one that's almost guaranteed to work on any video card, and it's not loaded on boot. If you want to use it, you have to run 'startx' from the command line. I'd rather it weren't there at all, but I'll accept the minimal cost of a few megabytes of space. I can spare it, on a 40 GB drive. Installation number 5, run over the network to avoid YAST2 (there doesn't seem to be any option to avoid it, curse them) succeeded perfectly. All that remains is to install the printer drivers, get the printer visible from both Windows and Linux, configure SAMBA and NFS, create and begin a backup scheme...

You know, it was a bad time to be muzzled for me. The Allchin thing... I could have had a lot of fun on this site with the whole Allchin Affair. (And even more fun with Stallman's response to it. Wow! Somebody get me a pin, that ego needs deflating FAST...) The various times Bush - sorry, Shrub - did something stupid. Well, at least with that one I can be confident of plenty of opportunities in the future. And you can bet I'll take advantage of them. <G>

In other news; no, I haven't forgotten about Conestoga. I'm digging through my old reference books and web sites, hoping to be able to start with something a little more than a vague "let's go to the Moon." I don't have everything needed right here, but I've got enough to start putting up some numbers and some serious thoughts and intentions, and that's what I'm going to do. There'll be a separate site for Conestoga, and I'll announce it when it's launched. In the meantime, check out the new Netwidows page; I rather like it. Cleaner design, with a good compromise among all the different personalities there. And no, I'm not saying that just because I'm married to the webmistress over there.

In closing tonight, I wanted to share something that ties in with Conestoga, and computers, and life in general. It's something I've thought many, many times myself in my short career as an engineer, as a computer geek, and before - and since. I've never found the right words for it, but I couldn't have put it better than this. Adapted directly from "A Rhapsody on Art and Engineering" at Nomadic Research Labs:

It is essential, when designing a complex system, to spend some relaxed time fantasizing about what it will be like when it's finished. After all, this is what drives the process of engineering: at some level between rigorous and fanciful, an image of the finished product must be held in the mind, savored, and examined from all sides. Only after this playful interlude (which, to a manager, may be disturbingly indistinguishable from unproductive wall-staring) can decomposition of the design into subsystems, tasks, and packaging make any sense.

Trying to shortcut this by starting on Day One with formal design methodologies can have the catastrophic effect of committing one to an ill-defined goal state, whereupon the end result is shaped more by design tools than the supposed objective. That's why so many products seem malformed, patched, and otherwise inelegant... the industry loves formal tools, and generally looks askance upon such frivolous notions as approaching product design as a delicate blend of art and engineering. The exceptions, when they occur, are a joy to use. The rest simply miss the point, no matter how stylish their exterior... or how sophisticated the underlying technology.

This complements something I've long noted and believed - that the best engineers are not the "pure math" geeks, no matter how fast they might be with a slide rule or calculator. To be an engineer is to me to be one of a select club of individuals, people who take raw, uncut knowledge and the roughest of materials to create magic. That's not a faster car or a larger aircraft - that's the stuff of dreams, cut from whole cloth and shaped into reality. Too many times in college I watched as professors and students alike strove to make all the designs the same, forced everyone into the same methods and the same rules. Is it any wonder that all cars look the same, such that you are forced to look for minor glued-on bits to distinguish a Ford from a Honda? Technology *is* magic, in the sense that some machines have a holy sort of grace about them, that indefinable something that makes us simply look at it at say "Wow." It's the difference between a motorcycle and a Harley, the difference between a bridge and the Golden Gate. Engineering is not science. Engineering is not art. Engineering is a fusion of the two, a strange and wonderful world where everything is both and neither at the same time. A world that's being destroyed by computer-aided design and manufacturing, by management techniques and the bottom line and liability lawsuits.

And that is why I'm a sysadmin.

Good night, all.

Thursday, March 29 - Time For Percussive Maintenance

Yep. That's right, BRIGID is broken again. Worked fine for several hours, until I rebooted the machine. Got the dreaded "LI" incomplete start, and locked it up tighter than... well, tight.

So now I'm experimenting with other settings, without much success. Twice now installations have failed to install LILO correctly; running LILO after configuring the profiles (which installs LILO in the MBR and prepares the system to boot) reported "Fatal: open /boot/initrd_24: no such file or directory". Needless to say, the system doesn't boot when LILO doesn't install properly. I've gone back to the 2.2.18 kernel; I don't really need any of the features of the 2.4 kernel series on this box, and so there's no reason to push my luck. Didn't help. OK, let's see... I had this box working before. What was different? Well, I had a different hard drive in the box. All right, pull it out and install it in the system. No difference. OK, maybe the RAM went bad. Swap it out, one stick at a time (there are only two). Nope. Same problem. Processor? Well, but it runs everything fine until I reboot. OK, we're back to the drive. Bad MBR? On two separate drives? Don't think so. Try installing LILO on a floppy anyway, just in case. Nope. Still no good. Very, very odd. I can't make myself believe it's SuSE; I've got two other servers and a laptop all running SuSE 7.1 with no problems. Still, if I can't make this work tonight then I think I'll pull out the Mandrake cds. All right, it's an older kernel and older software - so what? The only thing that'll really be affected is NFS 2.0 protocol instead of 3.0, which won't really matter with less than five machines connecting anyway. BRIGID is a completely internal box; the only possibility for outside access is that I might put a database on BRIGID linked from a site on THOR. Even there, I'm more inclined to simply add storage to THOR.

And the latest attempt, while promising, fails as well. Ah well. We have a rainy weekend coming up, perhaps I'll put it to use by beating it into submission. Or perhaps I'll be the one to be beaten. <G> Either way, it should be interesting.

With apologies to Mary Schmich of the Chicago Tribune...

Use a good chair.

If I could offer only one tip for your career, a good chair would be it. The long-term benefits of a good chair have been proven by legions of sysadmins, whereas the rest of of my advice has no basis more reliable than my own considerable ego. I will dispense this advice now.

Enjoy the power and beauty of the command line. Oh, never mind. You will not understand the power and beauty of the command line until you've long forgotten how. But trust me, in two years, when you're trying to reinstall Windows 2002 for the forty-third time, you'll look back and remember how wonderful it was when everything was on the command line. It is not as clumsy as you imagine.

Don't worry about the anti-trust case. Or worry, but know that worrying is about as effective as trying to hurry a disk format. Software comes and software goes, but the real troubles in your life are apt to be the things that never crossed your worried mind, like remembering where the license keys are at 4 p.m. with your boss watching.

Do one thing every day that scares you. Invite the users into the server room.

Backup.

Don't be reckless with other people's data. Don't put up with other people being reckless with yours.

RTFM.

Don't waste your time on Solitaire. Sometimes you win, sometimes you lose, and sometimes your boss walks in and you kill the game with two cards left. The game is long and, in the end, it's the only perfect program Microsoft ever made.

Remember the compliments you receive. Forget the trouble tickets. Just remember to assign them to someone else first.

Keep your reference letters. Throw away your MCSE.

Drink caffeine.

Don't feel bad if you can't think of a good screen name. The best and most interesting screen names are the ones you never thought of, the kind of thing you start using on a whim and the next thing you know more people know you by that name than your real one.

Use a good keyboard. Be kind to your wrists. You'll miss them when they're gone.

Maybe you'll learn Unix. Maybe you won't. Maybe you'll be an MCSE, or maybe you'll write the next killer operating system. Whatever, don't think you're too 'leet, nor flame others either. Your choices are at least half chance. So are everyone else's.

Enjoy your operating system. Use it every way you can. Don't be afraid of it or of what other people think of it. It's potentially the greatest friend you'll ever have.

Compile, even if you have nowhere to do it but your home directory.

Read the README files. Don't always follow them.

Do not read vendor-published magazines. They'll only make you feel ugly.

Get to know your senior admins. You never know when they'll be gone for good. Be nice to your junior admins. They're the ones most likely to cover for you.

Understand that systems come and go, but software lives forever. Learn to bridge the gaps in processor speeds and system architecture, as you get older you'll need it to impress the young snots.

Adminstrate Solaris once, but quit before it makes you bitter. Administrate Windows once, but quit before it makes you soft. Go slumming with a Mac.

Accept certain inalienable truths: systems will grow faster. Users will become disrespectful. You will get old. And when you do, you'll fantasize about when you were young, when systems were simple, users knew their place, and newbies respected sysops.

Respect the sysops.

Support yourself. Maybe you'll have a service contract. Maybe you'll have a warranty. But you never know when either one might run out.

Don't mess too much with your partitions or by the time they're backed up they'll already be full.

Be wary of who's upgrades you buy, but be patient with those who sell it. Upgrades are like a form of nostalgia, for the ten minutes after installation when you imagine the system is as perfect as you imagined it was when you installed it.

But trust me on the chair.