Monday, November 26, 2007

Entertaining For Geeks Only

I blogged previously about how on the way to Alaska for Thanksgiving, we found the secondary IT infrastructure around travel had become so nearly perfected. On the way back, we found a marvelous counter example.

We boarded a Delta Airlines 757 in Seattle, bound for Atlanta. This was a very modern plane equipped with a personal video screen for each passenger embedded in the back of the seat in front of you. The screen could be used for movies, or TV, or games, or to track altitude and speed statistics for the trip.

I wasn't interested in any of that stuff, but I do like fiddling with computers, so I decided to try our the free trivia video game they offered. I selected the game and pushed the Play button. A message appeared saying, "Please wait while the game is loaded." I waited and waited for about a minute. Then, to my surprise, the computer rebooted and I found myself staring at the boot script log on an 80x24 text screen. I hadn't seen one of those things for years.

The reboot seemed to take forever, much longer than it takes my laptop to boot Windows XP. Then I tried again. Surprise: selecting that game crashed the computer again. This time I looked more closely at the boot log. To my surprise, it wasn't running Windows, it was a I386 Intel machine running Red Hat Linux. Like any geek, I became interested in this glaring error so I started to investigate. Timing with my stopwatch, I determined that it took this computer between 185 and 200 seconds to reboot -- that's much slower than Windows. I wondered if my repeated reboots would affect other passengers but that didn't seem to be the case. I tried a race. I selected the game on my screen and on my wife's screen in the next seat. I could see that the two machines had the same error and they rebooted in approximately, but not identically the same time.

Looking at the boot log, I could see that there were a dozen or more errors reported during booting. Mounts didn't work, shares could not be found, services failed to initialize. It really looked like garbage.

After many tries, I figured out that the play button had three outcomes. It might work correctly (2 times out of 64 tries), it might fail and return to the previous screen (about 30 times out of 64 tries) or it might cause a reboot (about 32 times out of 64 tries). Further, the reboots always occurred 55 seconds after pushing play. I think the culprit was a non responding file server someplace plus a watchdog timer set to reboot after 55 seconds of non response.

Several times, the reboots themselves failed, getting stuck at some point in the process for more than 55 seconds thus causing a new reboot. In the worst case the reboot failed 3 consecutive times, resulting in 11 minutes until a successful reboot. The other passengers walking by who saw my screen wondered what I was up to with the B&W text logs scrolling by. It felt like the old days with a minicomputer and an ADM3A dumb terminal.

I walked around the cabin peeking at the screens of other passengers. Many were happily watching movies without trouble. I did notice a half dozen of them playing the trivia game that crashed my machine. That means it was not the game, but my (and my wife's) computers that were at fault. Several others had a static error message on the screen with the word ERROR in big block letters. They might have been trying another game. I overheard one passenger complaining that he watched HBO but the screen wanted to charge him an additional $5 every time one program ended and another started. Other passengers did not have this problem, indicating that his software was different than others. There is no way to blame the lusers in these cases. These computers have no keyboards and only very limited touch screen selection options. There is little room for operator error.

What a piece of crap! It was incompetently administered. Evidently they also could not do a push install of software so that all seats had the same software. God only knows how this aircraft's software compared with other aircraft's copies.

I know that the open source community is fond of trashing Microsoft and Windows. They may be right that Linux at its best (or even at its average) are better than Windows. However, I think that both of them at their worst are worse than Windows at its worst. Windows at least gives one the option to reboot to DOS or to boot in Safe Mode if the primary reboot fails. Clearly, the intricacies of properly administering Linux are beyond the skills of even major players (I count a supplier to Delta Airlines a major player.)

To be fair, I've seen similar disasters happen with boot scripts on Macs and on the fabled VAX/VMS machines, but those happened nearly 20 years ago. It's shocking to see such bumbling in 2007.

How can they consider putting such overqalified, over complex stuff such as Linux in mass consumer items such as cell phones and GPS receivers. I have a Lowrance GPS/chart plotter in my boat and I use it (plus paper charts) for safety related navigation. I saw once a message box that indicates that it runs a version of Windows. It has never crashed, never misbehaved the way the Linux based TV screen did on that Delta flight.

I'll say it again, Windows at its worse is better than Linux at its worse or Macs at their worse.

No comments:

Post a Comment

Type your comments here.

Note: Only a member of this blog may post a comment.