Day Notes

A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books.


[Monday]  [Tuesday]   [Wednesday]  [Thursday]  [Friday]  [Saturday]  [Sunday]

[Preceding Week]   [Day Notes Home]   [Following Week]


Monday, October 12, 1998

I changed the way I do things today. It's too easy to get caught up in doing stuff on the web site to the exclusion of stuff that makes money. A couple of times, I've found myself at noon still working on the web site, and that won't do. So, starting today, I've decided to start updating the site in the afternoon rather than in the morning. As a matter of fact, I put a recurring appointment in Outlook to remind me to work on the site from 5:00 p.m. until 5:45 p.m. each weekday afternoon.

For today, at least, this scheme appears to be working. I got a lot more done on the book today (about 2,000 new words plus a bunch of re-write) than I've gotten done any day in the last couple of weeks or so. The deadlines are still reasonably far away, but I can see them on the horizon now, so I have to go heads down on the book. That may mean fewer (and shorter) web site updates over the next couple of months, but it can't be helped.

Reader mail from last night::

There seems to be some sort of issue dealing with the processor module retention hardware, apparently different for PII and Celeron.

Hmm. That's news to me, although I've not actually installed a Celeron, nor even seen one. However, the processor retention mechanism actually comes with the processor, at least in the case of the Pentium II. It snaps into matching holes on the PII system board. I suspect that, although the Celeron may use different retention hardware, it fits into those same holes. The system board I just installed in new thoth lists support for both the Pentium II and the Celeron, so I'm sure that must be the case.

?) Some PII boards give better performance than the "Celeron" targeted boards.

Well, I can believe that. The Celeron is Intel's "economy" processor, so it makes sense that system boards designed specifically for it would be less expensive and slower than equivalent boards for the PII. I'm not aware of any Celeron-only boards, however. I'm not sure how there could be, because the Celeron is just a Pentium II with different L2 cache arrangements. I suppose that someone could have designed a motherboard with L2 cache built-in for the first, cacheless Celerons, but those are dead products.

?) availability of 100MHz frontside bus on PII boards (?BX)

I'm not sure if this is a question, but the Intel 440BX chipset does indeed support the 100 MHz FSB. I'm using one right now (an Intel Seattle BX) in the system I'm writing this on, but it's using a 300 MHz PII and configured for 66 MHz FSB.

I'd expect a 300A to run circles around a 333 or 350 AMD, as much as I like the AMD chip, so is an issue for low cost systems

I'll take your word for it. I've never worked with AMD processors, at least not since I was using 286s.

Your Day Notes are turning out to be informative and ... well, keep it up. I'm enjoying them.


And a couple of more from today, one about the boot failure on the replicated drive when I was juggling NT boot drives around, and the other about Christopher Morley:

I have had problems with copying hard drives also.

This case sounds like that the drive mirrored to was jumpered for Master and reinstalled in the box. Also since it was a slave the disk partition was NOT marked as active. If it was still NOT active the drive will not be bootable.

I've had this happen several times and "try" and remember to check it every time now.

No, that was my first thought as well, but it was jumpered as slave in order to be mirrored. I rejumpered it as master/sole drive before I reinstalled it and tried to boot from it. I also checked the status of the partition, and it was marked active. In fact, I disabled and then re-enabled that flag just to make sure, and it still wouldn't boot. Thanks for the note.

Long time editor of the Saturday Evening Post, and long time president of the Baker Street irregulars, Morley is the only author I know of whose work was so popular that someone collected the introductions he wrote to other people's books, and published a collection of them.... His poetry is so far out of our time, that it is mostly unreadable now, but some of his books, especially the two Parnassas ones, and perhaps Kathleen, look to long stand the test.

I agree, and I'm glad my wife found him for me. I'd heard his name, of course, but never read any of his work. Thanks for your note.

Well, I actually cheated and started working on this about a quarter to five. It's now about a quarter after and I have several things to get done, including making sandwiches for my mother and me for dinner. Barbara is at some kind of workshop she left for straight from work. I guess the dogs deserve a walk as well...

Dinner time: This is very aggravating. I tried to publish the web site changes at about 5:15, but FrontPage refused to co-operate. It kept giving me a time out error on my remote web site, which was odd because I could access it with a web browser with no problem. That's another aggravating thing about FrontPage 98--trying to figure out what's going on. There's a status line at the bottom of the screen. When you publish, that status line changes to update what's going on--scanning the local web, scanning the remote web, uploading changed pages, etc. The trouble is, you have to leave that status line visible all the time, or isn't refreshed. That means you can't minimize FP98 while you're publishing. You just have to let it sit there, or you'll have no idea what's going on. Sometimes, it appears to be locked except for the status bar display. If that's missing, there's no way to tell if you're locked, or if FP98 is just doing it's thing.

I seem to remember that Microsoft bought FP originally. Didn't it used to be named Vermeer? It's pretty obvious that they didn't do a re-write of the entire thing. There are just too many departures from the standard Microsoft way of doing things. I understand that FrontPage 2000 is a total re-write, and will be looking forward to seeing it.

Tuesday, October 13, 1998

I've started having problems with FrontPage 98 getting my updates published to the web site. This all started yesterday afternoon when I attempted to publish and FP98 timed out with an error message that said I needed to talk to the web hosting company. I mailed my web provider and told them what had happened and what the message said. They mailed me back and said they couldn't find any problem in their logs, and that I should try again and if it failed give them the exact time. So I tried again, and then sent them this message:

Okay. The first time it blew up was at about 17:15 eastern time. It got to the last phase "processing web updates for" and then timed out. I figured I had a TCP/IP problem, excessive latency or whatever, so I tried again. Same result, but this time instead of just delivering the message about timing out, it displayed the one I quoted in my last message. I tried dialing in to a different ISP, thinking maybe BellSouth's pipe was clogged, but had the same results.

After I got your message, I tried publishing again at 21:00 eastern time. The process ran normally, if slower than usual, through 21:06, when it blew up during the "processing web updates..." phase. I immediately tried again at 21:07, and this one completed normally at 21:12. Something isn't right, but I don't know if it's on your servers or if the net is just clogged big time.

To which they responded:

The "processing web updates" phase runs a bunch of internal functions on your web - text indexing, hyperlink recalculation, etc... These WILL TIME OUT on large webs. It is a known problem - Microsoft knows about it. The only solution is to break large webs into smaller ones. Is your web site particularly large?

I don't know what "large" means, but I don't think my site is very large. I checked Properties on the folder that contains the local copy of this web site. It had 1,823 files (106 *.htm* files) in 22 subfolders, and occupied 4.99 MB. If that's too big for FP98 to handle, they need to put a label on it, "Good for Toy Web Sites Only". I can't believe a site this small is really beyond the capabilities of FP98.

I forwarded a copy of this mail exchange to Pournelle, who has an ongoing conversation right now with the FP product managers from Microsoft, and asked him to check into this. I checked my local copy of Pournelle's web. It has only 800 odd files, about the same number of subfolders, and totals over 7 MB. I suspect the differential in file count is due to the fact that I'm using FP themes and he's not. I could cut down on the number of files by changing my Navigation Bar Properties to use text rather than images, which would also make things load faster. Perhaps I'll do that. It'll be interesting to hear what Microsoft has to say about all this.

Afternoon: More Word weirdities. This morning, as I saved the chapter document I was working on, Word locked up tight. The status bar incremented out towards 100%, but when it got there, it never finished saving. I finally used Task Manager to kill the Word process, figuring I'd lose the few minutes of work I'd done since the last save. When I tried to call up the document, it refused to open in Word. Okay, I use a batch file at least once an hour that xcopies the changed files in my data time to a network drive. Time to revert to the version on the network drive. Nope, that one's hosed, too. Before I go back to the version I stored on the backup tape last Friday, I decided to try to retrieve a usable version from the recycle bin. Thank god for Norton Utilities for NT. I was able to retrieve a version a day or so old. So, I ended up losing a day or so worth of work, but it could have been much worse.

This version reads okay, but it's acting weird as well. I save in Word 6/95 format (and, yes, I have SP1 installed) to make it easier for editors who don't have Word97 to read my stuff. I just spent 15 minutes or so working on the document and then did a save. I seldom go more than 15 minutes without saving, and I'm glad I hewed to that rule this time. Word popped up a dialog box to tell me that it couldn't save the document because there was inadequate disk space or memory to do the conversion. Hmm. There's almost 2 GB of free disk space on the network drive I'm saving to, so that can't be it. I ran Task Manager to check on memory. It shows about 41 MB being used from a total of 256 MB, so that doesn't seem a likely cause either.

Fortunately, Word hadn't locked up, so I was able to do a Save-As, kill the annoying little wizard gremlin, and save the document in native Word97 format, which worked fine. I hate Word.

Dinnertime: I got the Word problems resolved, kind of, although I'm still not sure what's causing them. I also spent a half hour or so getting all my mail server-related stuff reconfigured so that Barbara can get mail here at home. I did have things setup so that everything  that was addressed to a account (with the exception of a couple special accounts) ended up being forwarded to my ISP account with Being a cheap author, I didn't want to pay (my web hosting company) five dollars a month for another POP account for Barbara. Nor did I want to pay three dollars a month for the same.

So, I changed all my forwards. Now, anything addressed to a account goes to my local POP mailbox on the server, with a couple of exceptions. Stuff addressed to is set to forward to the account, where she can POP it. Stuff for (my co-author, Craig Hunt) goes to his real dialup account, and stuff for (my friend Steve Tucker) goes to his dialup account. Everything else to goes to the one POP box from which I POP. I let Outlook 98 sort all the stuff to different addresses out locally. It looks like this all may work, and Barbara will finally have her own account that doesn't go to my mailbox.

This is about to drive me mad. After finishing the above paragraphs, I tried to publish them to the web server. Once again, during the final phase, FrontPage 98 blew up with a timeout error. This was at about 6:00 p.m. We've now had dinner and cleaned up, and it's about 7:00 p.m. I'm going to try this again.

Wednesday, October 14, 1998

Well, the update I tried at 7:00 p.m. gave a time-out error. I came back to try again at about 8:00 p.m. This time, everything appeared to proceed normally. After several minutes, FrontPage came back and told me that there were no newer pages to be published. I hit the web site, and sure enough, the 7:00 update appears to have succeeded, despite the timeout message. I'm beginning to hate FrontPage. I'm going to try publishing this now (about 8:15 a.m. Eastern time) to see if the time of day makes any difference. I never had any problems when I was publishing updates in the morning, and I've had nothing but problems trying to do so in the afternoon and evening. Probably coincidence, but worth a try.

Okay, I just (8:12 a.m.) published the updates and everything completed normally. I'm going to try again. Surely this can't be sensitive to time-of-day.

Well, I just (8:20 a.m.) published normally again. And speaking of times, I got the following mail:

Sometimes you mention the times that you publish pages but when I retrieve the pages I get an earlier version even after I refresh. Why are the times that show on the pages out of whack with the times you mention?

There's nothing wrong on your end. I use a FrontPage 98 bot to timestamp the pages. Every time I save a page, the bot automatically updates the date and time the page was last saved. The problem is, I'm Eastern time and my web server is Pacific Time. If I update and save a page at 8:00 a.m. my time, that page shows an 8:00 a.m. timestamp on the local copy of the web. When I publish that page to the web server, however, the timestamp appears as 5:00 a.m. because of the time difference. I actually noticed this problem very early, and attempted to fix it by using a timestamp format that included the timezone. FrontPage choked and died on that version of the bot, however, so I'm back to using the unadorned version.

My mother is sick. Actually, we're all sick. My mother, however, is 80 years old and spiking 102 degree temperatures every evening, a pretty good indicator of a urinary tract infection. Fortunately, Barbara convinced her this morning that it's time to go to the doctor. My mother may be 80 years old and frail, but she's sure stubborn. When I tell her something she doesn't want to hear, she ignores me. When Barbara tells her, she listens. She called and made a 1:00 p.m. appointment with the doctor. It doesn't look like I'll get much done today.

Afternoon: Back from the doctor. My mother has a lung infection of some sort, but the doctor didn't seem too concerned, even though she will be 80 years old next month. He was going to prescribe penicillin, but she's allergic to it. He ended up prescribing Ceftin, one pill twice a day for ten days. I was surprised when I picked up the prescription to find out that it cost $68, more than $3 per pill. How do old people on fixed incomes deal with stuff like this? That'd be the greater part of a week's income for many of them.

Now, it may be that this drug was the only good choice for my mother's infection, but I kind of doubt it. Many of the new and expensive drugs are little or no better than older drugs, but the older drugs, of course, don't receive heavy promotion by the drug companies. If I practiced medicine, particularly with older patients, I'd make the cost of the drugs I prescribed a major factor in my selection. If the expensive drug is really needed, fine. If it's not, why not prescribe the one that costs $4 instead of the one that costs $70?

I'm not a believer in big government, or in any government for that matter, but it seems to me that if we're going to have Medicare, it should cover prescriptions. Without such coverage, older people can be reduced to eating cat food in order to pay for drugs. If the government is concerned about ballooning costs, they should have an approved drug list that contains only the inexpensive and generic drugs, which would be fully covered. That would, at least, focus doctors' attention on the cost of these drugs.

Evening: Now that the newest box has become Barbara's main workstation, the Windows NT Server thoth, I'm now without a Windows 98 box. Old thoth (now gladly) has a 3.1 GB drive that's a single NTFS partition. About 2 GB of that is occupied, but much of that is stuff that's already been copied over to new thoth. So, because gladly is now only one of two BDCs for the TTGNET domain, it made sense to reconfigure it as a dual-boot Windows NT & Windows 98 system.

Instead of manually repartitioning and reformatting the disk into a 2 GB NTFS partition and a 1 GB Windows 98 partition, I wanted to save some time by repartitioning dynamically with Partition Magic. I put the Partition Magic 3.05 CD in the drive and Autorun fired up setup. PM informed me that it could be installed only to a FAT partition, but offered the option to create a bootable floppy. I chose that option, and PM created a boot floppy using Caldera OpenDOS 7.01.

When I rebooted with the floppy, it recognized the 3.1 GB partition without problem. Under options, I chose Resize Partition and told it to reduce the NTFS partition to 2.0 GB. PM ran for about 15 minutes, displaying status indicators all along the way. It then blew up with a fatal error. Hoping that it was something repairable, I got back to the PM main menu and chose the option to verify the contents of a partition. PM blew up almost immediately, displaying Critical Error #46: Seek Error. That sounds bad, like a real hardware error, although it's pretty coincidental that a hardware problem chose just this moment to rear its head.

I removed the floppy disk and tried letting the system boot to the hard drive. Up until now, there was never any problem booting. This time, Windows NT started its boot sequence, got to the blue screen portion and immediately rebooted itself. I tried powering down the system to make sure the drive had reset, but the same thing happened when I restarted.

This system sure is behaving as though there's a hardware problem with the hard disk. Just to be sure, I loaded the Western Digital Diagnostics, expecting them to report a physical problem with the drive. I ran a non-destructive surface test, which completed without error. When the WD diagnostics complete they require a reset, so I powered the system down. When I restarted it, the OS Loader 4.01 screen came up normally. When the boot sequence arrived at the blue screen portion, the system spontaneously rebooted yet again. It appears that PartitionMagic has screwed up my hard disk. More accurately, it appears to have damaged the NTFS partition, although the seek error that it insists on reporting seems indicative of a problem with the drive itself.

Instead of messing with it any more, I think I'll blow away the existing partition, create a 1 GB FAT and a 2 GB NTFS partition, reinstall Windows NT Server, reinstall Seagate Backup Exec, and then restore the tape I made Friday. Nothing much has changed on this box since then, and it will probably take less time to do this than it would to attempt to recover what's on there now.

Thursday, October 15, 1998

I got started on rebuilding gladly from the ground up last night. To start, I used the Western Digital Diagnostics utility to write zeroes to the entire hard disk. That done, I booted the Windows 98 Startup Disk and used it to run fdisk to create 1 GB partition that will contain Windows 98. The remaining 2.1 GB I'm keeping in reserve to contain Windows NT. After restarting the computer, I formatted C:, logged to the CD-ROM drive, and ran Windows 98 setup. Setup ran flawlessly, and within half an hour the system booted Windows 98 under the new computer name anubis. The only problem was that Setup didn't detect my network card, sound card, or tape drive.

First things first. To get connected to the network, I needed that network card to work. Taking the brute force approach, I fired up the Add New Hardware Wizard, chose the option to manually specify the new hardware from a list, and picked the 3Com 10/100 PCI Fast EtherLink card. Windows 98 blew up with a Rundll32 critical error. After cursing out Windows 98 for a while, it finally hit me that there's not a 3Com card installed in this machine. Well, duh.

I remembered that this is a combo (coax/UTP) card from Addtron, an AE-200JLN. I fired up the Add New Hardware Wizard, got to the list of network adapters, and scrolled down to Addtron. There were only two cards listed, and the closest match was an AE-200PNP. I knew this wasn't the PnP card, but I gave it a try anyway, and got another Rundll32 error. Time to hit the Addtron web site and find a Windows 98 driver. When I got there, they had a file with the latest drivers for the AE-200JL series cards, but when I downloaded and unzipped it, the latest file was dated 1996.

I was considering installing the Windows 95 driver, when it hit me that I had configured this card to run in emulation mode as a Novell NE2000 compatible. I fired up the Add New Hardware Wizard again, and scrolled down to Novell Ethernet adapters. The NE2000 wasn't listed, which seemed strange. Okay, the NE2000 was actually manufactured by Eagle, so I scrolled up to look at that manufacturer. The only one listed was an NE200T. No joy. Well, these cards were also sold as Novell Anthem NE2000s, so I looked for Anthem as a manufacturer. Not there. I finally scrolled down past the Novell entry and saw what I needed. Novell/Anthem was the next manufacturer entry, and the NE2000 was listed, along with NE2000 compatible. I chose that card, installed the driver, configured TCP/IP, rebooted, and everything worked fine.

With the Ethernet card installed and anubis talking to the network, the next step was to get the sound card working. This situation is a bit different from the network adapter. Windows 98 Setup detected the sound card (an Ensoniq Soundscape/Vivo PnP), but it didn't install it. When I fired up the Add New Hardware Wizard, I found out why. Windows 98 doesn't have a driver for it. This seems odd, because the Soundscape/Vivo is a pretty popular card. Gateway alone has sold about a million of them. I went over to the Ensoniq web site, where I found the latest drivers. This is a good company, a good web site, and a good product. I downloaded the latest drivers, installed them, rebooted Windows 98, and everything was working fine.

Everything, that is, except the tape drive. I'd installed Windows 98 backup during Setup, assuming that the Seagate TapeStor Travan TR-4 tape drive would be detected. When I fired up Windows 98 Backup, it informed me that no backup device was present. I ran the Add New Hardware Wizard, but could find no option to install the TR-4 drive. That's odd, because even Windows NT 4 supports it. You'd think a Windows version almost 2 years newer would also do so.

Okay, time to go look on the Seagate web site. I couldn't find any reference at all to Windows 98, but did find this little nugget about Windows 95:

"Seagate Technology does not provide "tape drivers" for use with the native backup applet under Win95. Seagate Technology will only support its tape devices operating under Win95 while using the Seagate software with which the drive came bundled. Seagate backup applications come with their own driver software for the Win95 environment. "

That seems pretty obnoxious. Disregarding the fact that the Win95/Win98 Backup applications are trimmed-down versions of Arcada/Seagate BackupExec, what happens if you need to use a Seagate drive to restore a backup tape made to another brand of drive using the native Win9x Backup application? Well, no great loss. I was interested in looking at the Win98 Backup application, but I plan to install Windows NT 4 in the other partition on this machine, so I can just run Arcada BackupExec from there.

At this point, Windows 98 is running fine. I'll install and configure Windows NT Server 4.0 on the other partition later.

Later: I installed Windows NT Server 4.0 to the free space on the 3.1 GB hard disk. The installation went completely normally until the first forced reboot. When the Windows NT boot sequence reached the blue screen, the system spontaneously rebooted and continued to do so in a loop. This is exactly the behavior I saw last night, and made me wonder if there really was something wrong with the system.

I checked BIOS Setup. I was expecting to find that Windows 98 had somehow changed the option for "Plug-and-Play Operating System" from No to Yes. That's not what I found, however. Under Boot Options, I found that something had changed the boot order from what it had been (floppy disk followed by hard disk) to CD-ROM, then hard disk, then floppy. There only possibility I can see is that Partition Magic and Caldera OpenDOS 7.0 somehow did this.

When I started to use Partition Magic last night, the system booted fine. I inserted the PM 3.0 CD and Autorun loaded the PM setup program. It informed me that it could only install to a FAT partition, and offered to create a boot disk. I chose that option, and ended up with a bootable PM floppy that used Caldera OpenDOS 7.0 to boot. The strange behavior, including the spontaneous reboots, started then. I was inclined to blame Windows 98, but it hadn't even touched the system at that point. I have a great deal of confidence in Partition Magic, and I can't believe that this is anything but an obscure bug. I'll be mailing my contact at PowerQuest shortly to ask him what's going on.

At any rate, I went in and changed the BIOS Setup Boot Options back to floppy disk, then hard disk. When I restarted the computer, Windows NT Setup continued normally. I installed Service Pack 3 and restarted the system. I'd forgotten about the tape drive, so I ran the Control Panel Tape applet. It recognized that the Seagate Travan TR-4 drive was installed, but needed to load the driver. It prompted me for the distribution CD, from which it loaded the driver. As a general rule, any time you load anything from the original distribution CD it's a good idea to reinstall SP3, so I did so. One of these days I'll actually do everything in the right order...

With all that done, the next step was to install BackupExec. I have a not-for-resale copy of Arcada BackupExec 6.1 on CD, so I stuck it in the drive and started the install. After that completed, I installed a patch file that updated the Arcada BackupExec 6.1 installation to Seagate BackupExec 6.11.204. Version 7 of BackupExec has been available, and I really need to contact Seagate about getting an eval copy. Still, 6.11 is more than good enough to do the job.

Each time I do the weekly full network backup, I put the most recent backup tape in Barbara's purse and retrieve last week's tape. That gives me a kind of field-expedient off-site backup, while still making it relatively easy to retrieve the most recent tape if I need it. She's at work now, and has the newest tape with her, so I'll wait until this evening to do the actual restore. What I want to do is grab all the stuff that used to be on old thoth's drive that I didn't copy to new thoth. More on that later.

I've been keeping an eye on the ranking for Windows NT TCP/IP Network Administration. I think the book actually shipped to the warehouses a couple of days ago, but Amazon is still showing it as not yet available. At any rate, it had been sitting in the 4,500 to 6,000 range pretty consistently for the last week or ten days. For the last couple of days, it's been holding in the very low four figure range--somewhere around 1,000 to 1,800. It looks like it should do well.

Lunchtime: I just tried publishing this without success. It didn't time out this time. I just got to the "listing pages in" phase and sat there doing nothing. I finally killed it. At this point, empirically it seems that there's no problem so long as I publish in the early morning here (before 9:00 a.m. Eastern) but problems arise after that. It may be that the servers become busy once business hours start on the west coast, or it may be an Internet throughput problem. I'm going to try publishing late this evening, 10:00 p.m. or so, to see if things work better then. Otherwise, I'll be back to publishing first thing every morning.

Friday, October 16, 1998

I was able to publish at about 2:10 p.m. Eastern Time, and never did get around to trying again late last night. That's because I was dealing with what I thought might turn out to be a very bad situation. It started yesterday afternoon when I got tired of the old thoth box sitting in the middle of my office floor with a monitor balanced precariously on top. Because it has the tape drive that I need to do my weekly network backup, I decided that it was time to get it installed in its planned position.

I juggled things around on my main desktop (a 6' 8" X 3' 0" door). From left to right, I now have the new box based on old thoth (named anubis under Windows 98 and hathor under Windows NT 4.0), sherlock, and kerby. I used an A-B-C-D video/mouse/keyboard switch box to share one monitor between anubis/hathor and sherlock, simply because there's not room for more. A 3/0 door sounds big, but it's not quite wide enough for three keyboards/mousepads, let alone three monitors, when you consider space occupied by the minitower system units, speakers, hub, switchbox, etc.

At any rate, once I got that cabled together, I restarted anubis/hathor as hathor (Windows NT 4.0). Barbara had arrived home earlier than usual, and I snatched the most recent backup tape from her purse and started to restore it. It didn't work. It should have worked, because I run a compare/verify pass on every backup. The problem was that BackupExec wanted the catalog files that it generated during the backup. These files were, of course, on the original thoth volume that ended up blown away during the problem with PartitionMagic.

Okay, that seemed reasonable. Since the catalog files were no longer on disk, I told BackupExec to catalog the media, figuring it'd either read the catalog direct from tape or read the tape's directory and use it to rebuild the catalog. After spinning the tape drive for quite a while, BackupExec blew up with a error message about encountering an unexpected problem while spacing to the end of the data, and returned a hardware failure error message. That didn't sound good. In fact, I had to restart Windows NT to stop the tape drive from spinning.

The manual and the Seagate web site didn't offer much help, except to suggest retensioning the tape and cleaning the heads, both of which I did. I tried reading the catalog again, with exactly the same results. By this point, I was beginning to believe that the tape really was bad, so I went back and tried the prior week's tape, and then the one before that, and then the one before that. Same results each time. This can't be. I do a compare/verify pass each time. How can all of these tapes be unreadable?

To make a very long story short, they weren't unreadable. Here's the problem. When you do a tape backup, the backup program can write a catalog of the files and folders it backs up to either or both of two places--the tape itself, or the hard disk from which the backup software is running. My copy of BackupExec is configured to write the catalogs to the hard disk only. I thought it was writing the catalog both to the hard disk and the tape. So, reasonably enough since the hard disk catalog was gone forever, I asked BackupExec to retrieve the catalog from the tape.

There was no catalog on the tape, and BackupExec's action was pretty unreasonable I think. Instead of simply telling me that there was no catalog on the tape and offering to create one by reindexing the contents of the tape, it simply blew up with a misleading hardware failure error message. That's not good program design, to put it mildly, particularly since the program itself is confusing. I should emphasize that I'm using an older version (6.11), and version 7.2 is out now. It looks a lot better (I believe the Windows 98 Backup app is based on it), but I haven't gotten an eval copy yet.

At any rate, once I figured out what was going on, I was able to locate and clear a check box called Use storage media-based catalogs. Once I did that, using the Catalog command performed as expected, scanning the contents of the tape and rebuilding the disk-based catalog from that information. I was able to restore my latest weekly tape successfully. I restored the contents of old thoth C: to a subdirectory on hathor D:, getting back all the non-data stuff that I didn't want to lose from the old thoth system disk. So, I had a scare, but it all turned out well.

Saturday, October 17, 1998

Robert Denn, my editor at O'Reilly, mailed me yesterday to tell me he was holding the first copy of Windows NT TCP/IP Network Administration in his hands, and that he'd get an author copy FedEx'd to me as soon as possible. That's always a great moment for an author--the arrival of the first copy of your work in its final form. is still showing it as "Not Yet Published." As a matter of fact, it's on their "Top 25 Not Yet Published List," which is encouraging. It should arrive in the warehouses very soon, if it's not there now.

If you'd like to buy a copy of the book, you can help me by ordering it from by clicking here. Amazon pays me a percentage of the purchase price directly if you order it from this link. I use these "commissions" to support this web site. Although ordering it this way helps me, it doesn't cost you any more than ordering it normally.

When I restored the important stuff from tape to hathor the other day, I was concerned about its disk drive. It started out with a 2 GB NTFS partition, and by the time the restore finished, there was only about 200MB free. I'd restored everything to the \oldthoth folder, so I compressed that folder to free up some disk space. I ended up with about 650 MB free after the compression, and deleting some other stuff finally left me with about 1.3 GB free. After all of this, the disk was thrashing pretty badly. If I'd set out to create a badly fragmented disk intentionally, I couldn't have done much better.

NT 4.0 doesn't have a built-in disk defragmenting utility. In fact, for years Microsoft claimed that NTFS wasn't subject to fragmentation. Wrong. It's less subject to fragmentation than toy filesystems like FAT, but it becomes fragmented nonetheless. There are two NTFS defragmenting utilities out there that I know about. One is included in the Norton Utilities for NT. I liked it when I first used it, but the more I use it, the less I like it. The Norton defragmenter locks up periodically for no apparent reason. It's done this on different disks, ranging from nearly full to nearly empty, and on different systems. It's never lost any data, but its lockups are disconcerting. I don't trust it.

The other NTFS defragger is Diskeeper from Executive Software, the folks who will be supplying the bundled defragmenting utility for Windows NT 5.0. You can download Diskeeper Lite to try it out. Diskeeper Lite doesn't have all the bells and whistles of the full commercial version. For example, it doesn't run in the background, can't defrag multiple volumes in one pass, and it doesn't defrag directories.  If you want to look at the full commercial product, which does do all those things, you can also download a 30-day eval version of the flagship version, Diskeeper 3.0, from the website.

I had Diskeeper Lite handy, so I used it to defrag the drive on hathor. When the status screen came up, hathor's drive was almost solid red, showing nearly 60% fragmentation. No wonder it was thrashing. Diskeeper Lite fixed that in short order. If you run NT, you owe it to yourself to take a look at Diskeeper. Diskeeper Lite is a fully functional NTFS defragger that is perfectly adequate for personal use on NT Workstation systems. If you run Windows NT Server (or if you want the extra bells and whistles for NT Workstation), buy Diskeeper 3.0. Highly recommended.

Enough of this. I need to get to work on the laundry and other household chores.

Got the laundry and other stuff done, and spent the afternoon reading the new book. It always looks different somehow when it's a real book as opposed to a document on the screen. I never print my manuscripts, so the first time I see my words on paper is usually when I get the final galleys before the book goes to press.

Barbara made homemade pizza for dinner and opened a bottle of wine to celebrate the new book. I don't drink much, but I guess one glass of wine a couple of times a year won't hurt me.

Sunday, October 18, 1998

Today is devoted to various things that need done around the house, errands, etc. We started with a trip to Lowe's (a home repair center) to pick up some stuff to build bookshelves in Barbara's office. We want to do floor-to-ceiling bookshelves on one twelve foot wall with a window in the center of it, and then wrap the shelves around to cover floor-to-ceiling on a four foot space between the corner and the window on the other wall. She wants five shelves, which will give her about sixty linear feet of shelving, not counting the floor below the bottom shelf.

We bought the same stuff I did for my office. It's from a British company named Spur. It comprises steel double-track verticals into which the shelf supports snap. All of the stuff has a white vitreous finish. We bought seven 84" verticals ($7.74 each), two 28" verticals ($2.97 each) for under the window, and 39 shelf supports ($1.74 each). With a box of #12 X 2.5" wood screws and tax, the total damage came to about $141. We also looked at their 1X10's for the shelving, but I didn't think much of it. The #2 and #3 stuff was over-priced garbage, and the clear stuff was ridiculously priced--$15 for a six foot 1X10 pine board. Barbara's going to talk to someone she knows who runs a lumber company. I'm sure we'll get better wood at a decent price.

I was kind of surprised at the checkout. We arrived with a cart laden with this stuff. I handed the lady at the checkout one of the long verticals and said, "we have seven of these." Same thing on the other stuff. I just gave her one sample to read the barcode from and told her how many we had. She rang it up using the numbers I gave her. We must look honest or something.


Read Faithful Unto Death by Caroline Graham last night. This book is the seventh in her series of Chief Inspector Barnaby mysteries, although the first one I've read. Her characters seem almost inevitably to recall those of Colin Dexter's Inspector Morse series, which she even mentions in the book. Unlike Morse, Barnaby is married. Like Morse, though, Barnaby is introspective, sharp with his subordinates, and listens to opera. Also like Morse, Barnaby's sergeant Troy is lower-middle class, although he is in every respect a less admirable person than Morse's Robbie. I think I'll ask my wife to get me the others in this series. I want to read a few more, but I don't think I'll start buying them.


Coming Soon (I hope)


Updated: 05 July 2002 08:14

Copyright 1998, 1999, 2000, 2001, 2002, 2003, 2004 by Robert Bruce Thompson. All Rights Reserved.