|
|
||
FAT32 Anomalies—Day 64© Copyright Darrell Anderson. I ran into a peculiar problem with FAT32 partitions and the Linux kernel. There is little discussion about this quirk and I suspect the primary reason is merely that most Windows users never notice or pay attention to file date stamps and archive attributes. Ignorance is bliss I suppose. I previously mentioned in another journal entry that when I ran Excel 97 or PowerPoint 97 in WINE that although I made no changes to a file, the file date stamp changed and when I booted back into NT4 I noticed the archive bit was set. For a while I thought this peculiar only to WINE because I had no such problem with Word 97. Several days after I first noticed this oddity I was trying to learn my way around Quanta Plus to support some simple HTML editing. I wish there existed a good O’Reilly book for Quanta Plus because the program Help Handbook covers only a few points and is obviously written for people who are professional web developers. Some of the material covered in the Handbook made me glassy-eyed and I had no idea what was being discussed although the author seemed quite excited. Quanta Plus could help a lot of part-time HTML people like me if there existed a good book or user manual. Anyway, after using Quanta Plus I noticed that some file date stamps were modified although I was certain I had not modified those files. So now I was suspicious of the Linux kernel. I did some surfing and tried several different fstab parameters but all to no avail. So the problem slipped away for a few days until the itch once again irritated me. This time I logged in as root and as I had done with my mortal user account, I opened Quanta Plus and fiddled with a few files without saving. No file date stamp changes. I then used the Konqueror file manager to copy a file to a new name. Konqueror preserved the file date stamp. I then tried a cp -p and saw the same result. Hmm. I logged in as mortal user and tried the same cp -p command. I received an error message of “cp preserving times. . .operation not permitted.” So, this was a file ownership problem. As with most users, I configured my fstab to mount my FAT32 partitions with root as the individual owner and users as the group owner. I then ran across a thread that the only way to resolve this problem is to mount the partitions with the user who needs ownership rights. Okay, I can do that on my system as I have only one “serious” user. I have a test login account, but I don’t care about that. So I edited fstab to mount my FAT32 partitions with uid=500 and gid=100. I then repeated all of my tests. Not once did a file date stamp change, nor any archive bits. Problem solved. Sort of. I find the solution a kludge. I wonder why the Linux kernel has so much trouble with FAT32 partitions. I realize that FAT32 is a brain dead file system, but why the file date stamp changes when my mortal user account is a member of the users group? The reason has something to do with individual ownership despite the group ownership. I don’t care to become a kernel guru at this stage of my life and I suppose for now I do have a solution, although a kludge. I had another glitch with my FAT32 partitions when I ran Scandisk. Again, only a minor error and the error was the same as before with a faulty directory pointer of some kind. Scandisk had no problem repairing the error, but this stuff is getting to bother me. Not once in seven years have I had a data problem with NTFS. I’m sorely tempted to restore all of my data partitions to NTFS and forsake sharing data through multi-booting. If I do that then my only option for sharing data files between Slackware and NT4 is with two computers and networking. I either could maintain my data partitions as local NTFS and share them, or I could move my data to ext3 partitions to my Slackware box and share that way. With the latter idea I would have to remap my NT4 drive letters to the new network partitions rather than mount them from the local disk. Any way I look I think I have to do something because all of these little FAT32 nuisances bug me. Deeply. I no longer trust the integrity of my data. So now I am more or less back to where I was several weeks ago. I need to rebuild my second box. After my partitioning fiasco I lost interest in doing that. I have to do some additional planning this time, however. My K6-III+ box hums rather nicely and I do not want to mess with that. However, that is the box to which I connect my wonderful Samsung 712N 1280 x 1024 LCD flat screen. My second box is connected to my old, and I mean old, Viewsonic 7 17-inch CRT, circa 1991. That monitor keeps on ticking but the difference between that picture and my LCD is like night and day. Yes, I could swap monitors, but my NT4 box is still my production system. Although I am comfortable with Slackware and KDE, that environment is not where I spend most of my computing day. Not yet. I could buy a second LCD monitor, but down the road I plan to use only one box. My only other option is to hold my breath with the FAT32 partitions. I definitely need to do some thinking about all of this! Finis. |
||