Firmware 4.40.23081

[JSWolf]

Originally Posted by compurandom

I'm getting database corruption with calibre on this firmware again, so I'm going back to rebooting before connecting via usb. I've had to reset the database twice this week, and the second corruption was pretty massive. I coudln't even repair by dumping the database.

Have you been going to the home screen before connecting to the computer? Give that a try before going down the road of rebooting.

[DNSB]

Originally Posted by compurandom

I'm getting database corruption with calibre on this firmware again, so I'm going back to rebooting before connecting via usb. I've had to reset the database twice this week, and the second corruption was pretty massive. I couldn't even repair by dumping the database.

Could I ask you to check if when you connect your LC to your computer without the reboot, if in the .kobo directory, you see files with names such as BookReader.sqlite-wal, BookReader.sqlite-shm, KoboReader.sqlite-wal and KoboReader.sqlite-shm. These are part of the write ahead logging/shared memory process and should be removed when the database is closed.

That was the issue with the Kobo firmware that was prone to database corruption in that the USB connection was established before the databases were closed.

[compurandom]

Originally Posted by DNSB

Could I ask you to check if when you connect your LC to your computer without the reboot, if in the .kobo directory, you see files with names such as BookReader.sqlite-wal, BookReader.sqlite-shm, KoboReader.sqlite-wal and KoboReader.sqlite-shm.

I have looked a few times and didn't see them, but then, I don't know when exactly the corruption occurred so I could have just missed it.

Does calibre check for those? Can we change it to or add a plugin that does?

[DNSB]

Originally Posted by compurandom

I have looked a few times and didn't see them, but then, I don't know when exactly the corruption occurred so I could have just missed it.

Does calibre check for those? Can we change it to or add a plugin that does?

calibre does not check for them. There are multiple ways to corrupt the database, this was just one that was visually obvious when you looked at the file system.

[compurandom]

Originally Posted by DNSB

calibre does not check for them. There are multiple ways to corrupt the database, this was just one that was visually obvious when you looked at the file system.

It would be super useful if calibre could check for the database scratch files before it opens the database, because after it opens the database, they are guaranteed to be there. So the only way to tell is to close calibre before I plug the device in.

[PeterT]

Unfortunately the damage is probably already done regardless of any access being done by calibre to the file system.

The presence of any of the wal / shm files means that processes have not been shutdown cleanly on the Linux side of the Kobo, before the file system has been shared via usb with the PC. I don't recall all the details but it's even possible that the shared file system has been remounted as read only on the Kobo side.

Sent from my Pixel 7a using Tapatalk

[compurandom]

Originally Posted by PeterT

Unfortunately the damage is probably already done regardless of any access being done by calibre to the file system.

That may be so, but continuing to open the database can't possibly make things better, and that's even more true if the linux side on the kobo wasn't actually made read only.

And not checking for this before opening the database makes this very difficult to detect and debug.

If I knew that the database was damaged, rather than let calibre damage it more by using it, I could reboot the kobo, attempt repair on the database before it gets worse, and/or restore the database from backup before it goes for another 2-3 days and gets further corrupted to the point where restoring the database isn't going to help more than just deleting the database and rebuilding it.

Without some kind of detection of this problem, I'm left with the best option of rebooting the kobo before every usb connection.

And no, going back to home screen doesn't fix this issue and never did. I'd say that maybe it makes it less likely, but really I think it's just waving a rubber chicken at the problem.

[JSWolf]

Originally Posted by compurandom

That may be so, but continuing to open the database can't possibly make things better, and that's even more true if the linux side on the kobo wasn't actually made read only.

And not checking for this before opening the database makes this very difficult to detect and debug.

If I knew that the database was damaged, rather than let calibre damage it more by using it, I could reboot the kobo, attempt repair on the database before it gets worse, and/or restore the database from backup before it goes for another 2-3 days and gets further corrupted to the point where restoring the database isn't going to help more than just deleting the database and rebuilding it.

Without some kind of detection of this problem, I'm left with the best option of rebooting the kobo before every usb connection.

And no, going back to home screen doesn't fix this issue and never did. I'd say that maybe it makes it less likely, but really I think it's just waving a rubber chicken at the problem.

I went back to the home screen on my Libra 2 and it always worked for me.

[compurandom]

Originally Posted by JSWolf

I went back to the home screen on my Libra 2 and it always worked for me.

If I go to the home screen, sometimes it works. If I go to the home screen and wave a rubber chicken at it too, then sometimes it works more.

[Cootey]

Originally Posted by compurandom

If I go to the home screen, sometimes it works. If I go to the home screen and wave a rubber chicken at it too, then sometimes it works more.

The importance of rubber chickens in technical troubleshooting is highly underrated.

[DrChiper]

... and never underestimate the usage of the enforcing hammer for fine-tuning...

[compurandom]

Originally Posted by DrChiper

... and never underestimate the usage of the enforcing hammer for fine-tuning...

Yes, waving hammers and screwdrivers in the general direction of computers frequently makes them decide to work suddenly.

[DrChiper]

Originally Posted by compurandom

Yes, waving hammers and screwdrivers in the general direction of computers frequently makes them decide to work suddenly.

That's quantum-mechanics in action for you ...