Diskette Puzzle

Last week the OS/2 Museum received a classic red NetWare box with all sorts of junk inside: PCI and ISA network cards (most Ethernet, one ArcNET), BNC cabling, one or two manuals, and over a 100 floppies, mostly NetWare but also a handful of 3Com driver disks.

There was a mix of 5.25″ and 3.5″ NetWare floppies, the 5.25″ ones in three original NetWare boxes but most of the 3.5″ disks just more or less loose in the big red crate. As I tried to organize the floppies, I quickly realized that it’s not that simple.

At a quick glance, there were floppies from several NetWare sets:

  • NetWare 2.2 on 5.25″ floppies
  • NetWare 3.11 on 5.25″ floppies
  • NetWare 3.11 on 3.5″ floppies, two sets
  • NetWare 3.12 on 3.5″ floppies

Now the difficulty with NetWare is that unlike, say, Microsoft or IBM, Novell didn’t just label all the disks in a box “NetWare 3.11”. There was in fact significant overlap and e.g. many disks were identical between NetWare 2.2 and 3.11, and later between 3.12 and 4.0. Related to that, NetWare didn’t refresh all disks for each update; only the disks that actually needed updating were changed. It was thus standard for a NetWare disk set to contain floppies with several different revisions.

Same or different?

That gets really complicated if you have a pile of disks and no easy way to tell which sets belong together. And it didn’t end there either: Novell shipped various add-on products with their systems, such as Macintosh clients, OS/2 Requesters, backup and mail software, and so on. And again, these were not labeled as belonging to a specific release, because they were to some extent independent.

In the big box there were, for example, four 3.5″ disks labeled WSGEN, part no. 136-000833-001 Rev. A. Interestingly, the labels didn’t all look the same and the floppies themselves had two different part numbers and copyright years. How does that match with three 3.5″ NetWare disk sets?

My best guess is that—since NetWare 3.12 did not come with a WSGEN floppy at all—the four 3.5″ WSGEN floppies belong to 5.25″ NetWare 3.11, 3.5″ NetWare 3.11 (two sets), and 5.25″ NetWare 2.2. That is a well informed guess based on other NetWare disk sets, but still only a guess.

Sometimes mass-duplicated floppies have markers, usually an odd-sized sector in the 81st track, that might make it possible to match floppies based on production runs, date of duplication, or some such data. Novell does not appear to have had any such markers, or at least nothing I could find in Kryofluxed images. The physical diskettes also looked in many cases identical, with identical “Made in Canada” labels and not other distinguishing information.

To top it off, there were also four 5.25″ WSGEN floppies in the box, with the same part number (153-0000049-001 Rev. A) but three different copyright years. I have to assume that they again belong to NetWare 2.2 and to the three NetWare 3.11 sets. To make system administrators’ life easier, NetWare shipped the WSGEN floppy in both 5.25″ and 3.5″ formats. The NetWare OS itself was delivered either on 5.25″ or 3.5″ media depending on the server, but the WSGEN floppies needed to be used on workstations which were expected to be a mix and typically not the same hardware as the server in any case.

Data from 1990, 1992, or 1993? Actually it’s from 1991…

In the end, I was able to more or less fully organize the floppies after looking at the contents of other NetWare disk sets. The big red box turned out to contain the following:

  • Complete NetWare 3.12 set on 3.5″ media with a 50-user license. This did not include online documentation, which I believe was an optional item and language specific. This would have cost almost $5,000 in 1993.
  • Two NetWare 3.11 Runtime sets on 3.5″ media with 1-user licenses. More on that below.
  • One nearly complete NetWare 3.11 set on 5.25″ media, sadly the only thing missing being the license disk. That’s a shame because I already have a 3.11 set on 5.25″ media that looks identical and the license disk is the only thing that would have added any real value.
  • One complete NetWare 2.2 set on 5.25″ media with a 5-user license. Back in 1991, that would have been the only sub-$1,000 NetWare package. This was welcome because I previously only had an incomplete NetWare 2.2 disk set.
  • OS/2 Requester 1.3 on 5.25″ media. This was almost certainly shipped with NetWare 3.11.
  • German NetWare Lite 1.1 on a 3.5″ disk.
  • Original Novell NE2000 5.25″ driver disk (LAN_DRV_107).
  • UnixWare Personal Edition 5.25″ boot floppies (set of 3), of all things.

There was also a 1988 NE2000 boot ROM from Novell in an unopened anti-static box, which I also duly “imaged” with an EPROM reader:

NE2000 boot ROM from Novell

As far as I can tell, the 1-user versions of NetWare 3.11 are “runtime licenses” that were not sold as the normal NetWare NOS (NetWare 3.11 started with 20-user licenses, and generally 5-user licenses were the smallest Novell offered) but rather sold as part of some other product, such as NetWare SAA. The system floppies do in fact say “NetWare Runtime v3.11”.

With some of the floppies, it’s truly impossible to figure out where they came from absent original packaging. For example the DOSUTIL-1 to DOSUTIL-4 floppies are identical between NetWare 2.2 and 3.11. As long as the disks are truly identical, it does not really matter. But when the disks actually look different (such as different font or different media part number), that causes complications when scanning or photographing the disks and uploading the pictures to, say, archive.org. I mean what if someone took the Internet seriously?

Only after I sorted the floppies into product sets in a (to my satisfaction) reasonable manner, I realized that those NetWare products are old enough that maybe Novell published the disk lists. And sure enough, I found just that in a 1992 Novell Support Encyclopedia issue:

NetWare v3.11 Diskette Listing

Not only was the list of floppies there, all the files with their sizes and time stamps were listed, too:

Contents of the LAN_DRV_001 disk

When everything was sorted, I was left with a floppy labeled NetWare Hub Services V2.0, which could have belonged with one of the NetWare 3.11 Runtime sets. There was also a five-disk set of AVM-branded NetWare MultiProtocol Router for ISDN, which may have come with the other set of NetWare 3.11 Runtime disks. Maybe. There is extremely little to be found about the NetWare Runtime kits.

There were over 180 floppies in the box, nearly all of them high density. Two were 3.5″ DD floppies in funny Macintosh CLV format that Kryoflux had trouble with, although I have no reason to think the problem was with the actual disks. Of the PC-formatted floppies, all except one read without errors. The only problem floppy was a 5.25″ HD disk that was a German CHIP magazine supplement, and it mystified me a bit: While one side of the disk read without errors, the other was almost completely unreadable (not recognized as formatted by Kryoflux) but there was no apparent physical damage to the disk. I don’t know if that floppy got too close to a magnet or what.

All in all, a pretty successful archiving run.

This entry was posted in Archiving, Floppies, NetWare. Bookmark the permalink.

20 Responses to Diskette Puzzle

  1. ths says:

    I was using Netware Lite for a very short time on a test setup with 2 machines, and for the first install I didn’t care to use the 2nd disk set, just installed both machines from the same floppies. I received a “license exceeded” error upon start. Used the 2nd disk set and things were fine then.

    Comparing the disks file- and byte-wise I noticed there were exactly 2 bytes different, and I assumed this were the license codes ;). Memory is weak but I seem to remember this was in server.exe.

    NWL was the first “peer to peer” network, and I cannot recall it gained a lot of attendance, especially after Windows for Workgroups 3.11.

  2. Michal Necasek says:

    I’d think the first peer-to-peer network was Artisoft’s LANtastic, except that’s not really true because the IBM PC Network Program was older than that (and it was a peer-to-peer network based on Microsoft’s network stack).

    The NWL serial check must have been taken from NetWare, that did the same thing. If two NetWare servers with the same serial number popped up on a LAN, they let you know about that pretty loudly.

    NetWare Lite 1.0 didn’t get great reviews, NW Lite 1.1 was supposed to be noticeably better but that was right around the time the first Windows for Workgroups showed up, and that very quickly took over.

  3. Richard Wells says:

    The Netware runtime was used for programs running Btrieve which might not want a full Netware server just to run the database. Netware even has documentation on it: http://www.novell.com/zh-cn/documentation/nw5/nw5/usserver/btrv_enu/data/h8dpvswp.html

  4. Michal Necasek says:

    Yes, that’s it… and they explain that even though only one “normal” connection is allowed, the runtime server will support any number of database connections and such.

    Here is something about the initial rollout of NetWare Runtime v3.11 from Feb ’92, although there is no pricing. And I think the “takes the file serving out of NetWare” bit is not strictly true, it was all there but with a single connection allowed it would have been impractical to use as a file server.

  5. John Elliott says:

    Don’t know if this would apply to Kryoflux, but some flux-level tools have trouble with Macintosh GCR floppies that’s specific to the make of drive. From the Deluxe Option Board manual:

    It has come to our attention that not all 3.5 inch IBM compatible drives can reliably read Macintosh diskettes using the Deluxe Option Board (DOB). This is due to the fact that drives made for IBM machines are designed to read and write in a particular format, MFM. Macintosh disks use a different format, GCR.The option board allows these drives designed to read and write in MFMformat to also read and write in GCR format. Not all drives are capable of doing this. There is nothing wrong with these drives. They do what they were designed to do. However, you can not use them with the option board to read and write Macintosh diskettes.

    The suggestion on the Vintage Computer forums was that this could be down to a bandpass filter within the drive.

  6. Michal Necasek says:

    That would make sense… what I noticed is that the outermost tracks (0-15, with 12 sectors per track) are almost unreadable while the rest is almost perfectly readable. And it really does not look like disk damage (in fact several different Mac-formatted disks behave exactly the same). And the drive has zero trouble with MFM floppies.

    Have you seen any suggestions about PC drive models that are more likely to work? It’s not high priority for me so I won’t buy some exotic drive, then again maybe nothing exotic is required, just the right model.

    Maybe I’ll just go through the drives I have and see if any of them works better. Right now I have a Sony MPF920-E hooked up to the Kryoflux.

  7. Chris M. says:

    According to Central Point’s documentation, the following are NOT recommended for 800k GCR floppies: all FDHD models of Alps, Mitsumi, Mitsubishi*, Panasonic

    The following are recommended: All FDDD/720k drives and FDHD drives from Citizen, TEAC, and Toshiba. Central Point officially recommended TEAC drives if purchasing.

    While Central Point makes no note of it, I suspect Sony drives should work fine too seeing that was Apple’s OEM for the FDHD SuperDrive in Macs.

    *The irony of Mitsubishi showing up on this list is that late model PowerMacs came with a Mitsubishi floppy drive as a cost cutting measure. Even being driven by an Apple SWIM controller, they had problems reading 800k GCR floppies!

  8. Michal Necasek says:

    As it happens, the three 3.5″ drives I could quickly find were all Teac classics, FD-235HF. I must have too many random 3.5″ drives built into various machines. Despite the same model the Teacs don’t look anywhere near identical.

    One of them wouldn’t read anything, one only read one side properly (I had thought the drives were both good!), but the third was the charm. Not only could it read both sides of MFM floppies, it also read those darned Apple CLV GCR disks! Many tanks for the hints John and Chris.

    Some kind of bandpass filter in the drive makes sense, and it would definitely affect Kryoflux et al. since it would work in the analog signal domain, before the signal hits the drive cable. Learn something new every day…

    I did see that the FD-235HF drive that only read one side properly was able to read a GCR disk. So a good chance it’s a general property of the model. Ironically, the three Teac drives were made in 2000, 1996, and 1994, and the oldest of them was the only one working.

  9. Julien Oster says:

    A bandpass filter would indeed make sense, but also potentially open up the possibility to modify the drives’ filters to read more disks.

    I’m an amateur in designing analog filters, but if I ever come across the problem, I’ll gladly try.

  10. Michal Necasek says:

    I have a suspicion that the filter might be inside a chip, so there’s nothing to modify. But I could easily be wrong.

  11. Michal Necasek says:

    I’m told a filter is actually unlikely, but AGC going haywire sounds plausible.

    I tried to find out what RPMs those Apple CLV drives actually used, but got nothing. I’m guessing that the outer rim spins at a lot less than 300 RPM in the Apple drives and when that runs under the head of a PC drive, it might create much stronger signal due to the increased velocity. But I have no idea if Apple ran it at 280 RPM or 150 RPM or what.

  12. Richard Wells says:

    For some information about the Mac RPMs, look for a file called Apple 400k Floppy Disk Drive – 6990285A It’s a PDF. Appendix B (page 33) shows that there are two different Mac speeds. Speed 1 ranges from 394 RPM to 590 RPM while Speed 2 ranges from 402 RPM to 603 RPM. Speed 1 requires a data transfer rate of 489.6 kbps while Speed 2 uses the more normal 500 kbps.

    The document also has interleave and other information necessary to accurately process disks.

    I haven’t found a matching reference specific to the 800K implementation. Hopefully, no changes were made beyond using a second side.

  13. Michal Necasek says:

    Cool, thanks for digging that up. That’s actually interesting because even the lowest speed (394 RPM) was significantly higher than the 300 RPM of PC floppies, and the top speed around 600 RPM was twice as much.

    I should note that Speed 2 uses the more normal 500 kbps… to put 800K on a floppy, whereas on PCs you get 1.44M (or up to ~1.8M) with 500 kbps. But that’s logical if the Mac floppies rotated so much faster.

  14. Chris M. says:

    800k disks were just double sided 400k, so the drive speeds are the same. The early 3.5″ full height 400k drives that came in the Mac 128k and 512k were pretty loud, and you could clearly hear the spindle motor change speeds depending on what track of the disk the drive was accessing. These drives also needs a PWM signal from the host machine to control their speed.

    The later half height 800k mechs were pretty quiet, but you can tell each track had differing amounts of data on them. Fast copy programs that hit the IWM directly can do a high speed linear block copy and the time between head seeks changes as it goes out to the end of the disk.

  15. Richard Wells says:

    A couple of side notes.

    The Sony OA-D32 manual of Nov 1983 shows that the design ran at 600 RPM with planned transfer rates of 250 kbps or 500 kbps. There is a bit of an oddity with calling 250 kbps single density and 500 kbps double density. The V model is double sided; the W model single sided. This model does have a 300 RPM jumper but Apple would have already had 6 months of prototypes with the 600 RPM design.

    Two PDF over at Bitsavers that deserve looking at are
    Sony Microdrive Interface from May 1983 which lists a different set of rotations going from 363 RPM to 590 RPM. There may have been other slightly different variable rotations designed; I vaguely remember that something caused the very early Macs to fail to read 400K floppies written by later Macs.
    and
    Mac v IIgs sectors which covers the need for adding 30 nybles between sectors to keep the IIgs matching the timing of the Mac with a hint that there could be mixed format disks which could throw off all standardized calculations.

  16. Michal Necasek says:

    I remember reading somewhere how Sony had 600 RPM drives and how it’s a shame they were dropped because they should have been more reliable (higher velocity -> stronger EM fields). I guess that is what Apple actually used.

    The “CLV” designation of the Apple drives is a bit of a lie because they only have five “zones” but it still does something, and the RPM difference between center and rim is quite big.

  17. Chris M. says:

    Luckily the physical formats don’t differ between the IIgs and Mac, just different file systems (HFS vs. ProDOS).

  18. Richard Wells says:

    The physical formats are different; the sector gap is different. Having a different sector gap is one of the ways to get an Atari ST MFM FAT-12 disk format that no Intel/NEC floppy controller read. Apple put a lot of effort into being able to handle the variations that the IWM could achieve until the IIfx where the choice of IO processors meant some disks were no longer readable.

    The mechanical requirements of the drive were amazing. To overwrite a sector created on a Mac with a roughly 5% shorter sector followed by increased padding without writing over the beginning of the next sector requires all drives to rotate at very close to the same rate. Compare that to the usual floppy standard which accepts rotation rates that could be 10% off.

    The default format for 400K Mac floppies was MFS until Apple dropped MFS support.

  19. Vikki McDonough says:

    >Original Novell NE2000 .525″ driver disk

    >.525″

    Typo?

    Also, if the magazine-supplement floppy had gotten close enough to a magnet to corrupt one side that badly, wouldn’t that also corrupt much of the other side as well?

  20. Michal Necasek says:

    Fixed the typo, thanks. The disk was definitely bigger than half an inch.

    As for the floppy, I really don’t know. My experiments with ferrite magnets years ago indicated that one has to get surprisingly close to make much of an impact. It’s also possible that the drive head was dirty, although that usually causes problems with following disks too, and that wasn’t the case with this floppy, plus I don’t think there were the tell-tale scratches on the floppy surface.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.