Fat 32 disc problem

I use Safestore to backup to an external FAT32 disc, connected to the ARMX6 via usb.

Recently Safestore has been reporting “file not found”, and on investigating I find that though the source is there, the destination directory is faulty.

If I double-click the destination directory the hourglass appears and the machine freezes. No response to control-break or alt-break, and waiting half an hour doesn’t clear the hourglass either. No keys or mouse clicks work.

Trying to delete the directory reports file errors for something which doesn’t appear to be a valid file name within the directory, and retry locks up the computer again.

I don;’t want to format the disc, because the rest of the backup is valid. Any suggestions?

I tried plugging the disc into a Windows 10 machine. There the directory opens, and appears full of files, each of whose name seems to be part of an HTML page, and most if not all are marked as “hidden”. Attempting to delete them from there produces “not responding” from the Windows filer.

Does DiscKnight handle FAT32 discs? [Edit] Just tried – it doesn’t offer FAT32 as a file system.

Any suggestions?

1. Copy as much as you can off the failing drive onto another drive.
2. On Windows, run chkdsk /f on the drive.
3. Back on RISC OS, see if you can now retrieve more files from the drive.
4. Reformat (or replace) the drive.
5. Reinstate the contents on the new(ly-formatted) drive.

As for what could have gone wrong, I have a few thoughts:

• The drive may have developed a defect
• You may have a wonky PSU that is causing brownouts when the drive is operating
• A previously-undiscovered bug in Fat32FS could conceivably have messed up the directory contents

There may be other reasons I haven’t thought of.

Thanks Stuart.

There may be other reasons I haven’t thought of.

One possibility relates to a long-standing issue I have. The ARMX6, an RPC, a Windows laptop and a Windows2003 server are connected to a 4-way KVM. Due to two of the above being VGA, it’s a VGA beast using two HDMI=VGA adapters.

The issue is that when switching TO the ARMX6, the mouse and keyboard sometimes don’t connect. This occurs almost every time if I move the mouse while waiting for the monitor to sync and display the picture. (which takes 10 seconds in this case.)

Switching away and back a few times will correct the problem in 90% of cases. In the remainder, the only remedy is the power button.

I have tried using VNC to connect to the ARMX6 in this situation, and some times it works, sometimes reports “no response”.

Yesterday I noticed that the screen picture included the status window from Safestore, which was also stalled. That was one of the times I had to power off.

It seems as though the USB stack is locking up while trying to deal with the “new” USB devices from the KVM.

That power-off while Safestore was stalled might be one way the directory got corrupted. (Looking at the file names, it looks to me as though a block of memory containing the contents of an HTML file became mapped onto a directory.)

I find it interesting that Safestore is the only thing on RISC OS that can attempt to look into that directory without getting locked up itself.

It looks as though I will need to format the drive and re-run the backup. That will be the quickest and safest way to recover, albeit taking 2 – 3 hours to run.

I tried chkdsk /f, which produced the following:

The first “unrecoverable error” is the folder I was having problems with.

Now I have the choice of deleting all the folders converted to files, and re-running the backup, or formatting and rerunning the backup. I think the latter, but I might see what it’s actually produced first.

===========

Microsoft Windows [Version 10.0.19044.2075]
© Microsoft Corporation. All rights reserved.

C:\Users\Alan.LAPTOP-3>chkdsk /f d:
The type of the file system is FAT32.
Volume Backup created 13/03/2021 21:59
Volume Serial Number is 1D1A-151D
Windows is verifying files and folders…
\Disc256\Disc256 to external disc\Data\TransientFiles\Default\2021.02.22\2021.02.22\2020.11.27\Cardington_Sept_-_Simply_Slalom_Export Invalid time stamp.

There was quite a lot more pasted in – but it looks as thgough textile has truncated the report.

The first “unrecoverable error” is the folder I was having problems with.

If you can find a copy of Hirren’s Boot CD, burn that and boot from it. It’ll have different tools (perhaps with the same name, confusingly) that run under Linux or some sort of DOS.

I find Microsoft’s tools are often worse than useless at coping with problems in their own damn filing system. Many years ago when DOSFS was only capable of dealing with 2GB media, the resultant mess that it made of a 4GB USB stick actually caused XP to bluescreen as vfat.vfd (or something like that? keeled over dead).

or formatting and rerunning the backup.

Do you have something, maybe for Windows, that can test the drive? Is SMART data available?
It’s probably a high level fault (gibberish written to the map or something) rather than a low level (hardware) fault, but can’t hurt to be sure.

but it looks as though textile has truncated the report.

Looks like you were starting to include HTML, that might have upset it a little. ;)

Would FATfsck, from the same site as FAT32FS, be of use in this scenario?

I’ve now formatted the drive, using FAT32form, and re-run the backup. Three hours later it’s done.

As a result there’s little left to test. I’ve written 50GB to an 80GB disc without any reported errors, so most at least of the hardware is working.