Glo Is there any way to avoid bootloops on adding comics (CBR/CBZ)?

[Stremon]

I am using at lot my kobo glo to read mangas in comics (CBR/CBZ) format, but everytime that the reader struggles to read a file, which happens like half of the time, it hangs at media scanning and/or gets stuck in an infinite bootloops, until I remove the micro SD Card or hard reset.
Sometime it even does itself a factory reset... (happened to me like 4 times in 2 months)

I tried many methods and program to try avoiding this bug, but nothing seems to work well.
It happens the same way no matter if I put the files on internal or external memory.

I also changed the internal micro SD card by a good quality card, but it other than being slightly faster, it doesn't change anything.

I saw somewhere that a big capacity micro SD card can cause such bug, I try many different storage space (2Go, 4Go, 8Go, 16Go, 32Go), and indeed the biggest micro SD card have more problems than the smaller ones.
But no matter which one I use, it always does the same bug for the same cbz/cbr files.

Is there any way to fix this bug, or anything to modify in my cbz/cbr files to prevent it? Or even a clue of what might cause this bug?

Edit:
The problems came mainly from a lot of thumbs.db files (hidden windows explorer files created for thumbnail) and unsupported characters filenames (special characters like french letters with accent) inside the compressed files.
Refer to my last post for more informations.

[davidfor]

I haven't heard of issues with processing CBR and CBZ files. I don't read mangas or comics, but the few I have tested had no issues. As such, the following are some thoughts and questions that hopefully will lead to a fix or workaround.

I'm not sure what you mean by "the reader struggles to read a file". Are you meaning during the initial processing or when you open the manga to read it?

Assuming it is during the processing, how many manga are you processing at a time and how big are they? When CBR/CBZ are processed, the names of the files inside them are extracted and put into the database. As the format is a lot simpler than an epub, the only things I can think to go wrong is if the manga was very big with lots of images, or the internal file names were something that the firmware couldn't handle.

After the factory reset, how are you putting the manga back on the device? Are you adding a few and then adding more later. And then it goes into the loop? Or are you putting them all back on and it works OK, but later does the processing loop?

The factory reset is probably a built-in safety feature. Each time the device restarts, it checks to see if nickel (the UI) has started before. If several reboots have been done without nickel starting successfully, it will prompt to do a factory reset. With the continual restart you describe, this has probably happened a few times.

And something missing from your report, what firmware version are you running? Did this ever work and if so, which firmware was that?

[Stremon]

Thank you for your answer

Originally Posted by davidfor

I'm not sure what you mean by "the reader struggles to read a file". Are you meaning during the initial processing or when you open the manga to read it?

Mostly during the initial process, on some file the device just hangs on the progress bar and I have to reset it (it hangs again on the same point if I don't remove the file it can't go further), and on some other it scans correctly but does a bootlop right after starting nickel... Sometime it also crash, reset, or factory reset randomly, but it is still rare.

Originally Posted by davidfor

Assuming it is during the processing, how many manga are you processing at a time and how big are they? When CBR/CBZ are processed, the names of the files inside them are extracted and put into the database. As the format is a lot simpler than an epub, the only things I can think to go wrong is if the manga was very big with lots of images, or the internal file names were something that the firmware couldn't handle.

Honestly, it depends, many different sizes and numbers. But luckily this bug is not random, it almost always happens on the same files, no matter how many I put at the same time.

Originally Posted by davidfor

After the factory reset, how are you putting the manga back on the device? Are you adding a few and then adding more later. And then it goes into the loop? Or are you putting them all back on and it works OK, but later does the processing loop?

actually I always put the cbr/cbz files in the external microSD card, it is more convenient. Since it's not random, when it crashes I just remove the last files I added and it usually works.

Originally Posted by davidfor

And something missing from your report, what firmware version are you running? Did this ever work and if so, which firmware was that?

Ah sorry I forgot to mention that, it is 3.12.0

From everything I tried, and from what you said, indeed it seems that the device has some problem reading some very specific files.
My problem is I don't have any idea of what causes the problem.
It happens really often, and everytime I try modifying the file manually, remove some big images, renaming them... sometime it works, sometime it doesn't... but I have no idea why most of the time.
Sometime it simply works by uncompressing the file and recompressing it again, without changing anything else!!

Other information, it doesn't seem to be kobo glo or last firmware problem only, because if I remember correctly it was sometime doing the same on my good old Kobo touch too. But all the other devices/softwares I tested with those files don't have any issue.

[PeterT]

Two ideas; one is to check in the .kobo directory for the presence of any stack__.log files which are produced when a process dies on the device. If so, you need to get those to Kobo.

The other is to provide a copy of one of the failing CBRs to Kobo.

[Ken Maltby]

Koreader can read cbz files (including cbr files renamed with the cbz extension) and it can read them from a "hidden" folder, so you can keep those files away from the Kobo/Nickel processing that is causing your problems. I also suspect that the database can be corrupted dealing with files it rejects.

Luck;
Ken

[PeterT]

I've never seen ANY link between corrupted databases and rejected files.

[davidfor]

I could see the database being corrupted if the bad files caused a crash and reboot while the database was being written, but in the cases where the files are not processed, I don't think so. The firmware seems to process all the files and the database updates happen at the end in one go.

As to the crash, PeterT's suggestion is good. Look for some stack logs and supply a file that causes the problem. Kobo are the one who need to see it, but I'm curious to look as well. Ideally supplying a link to a freely available manga would be the best. But, putting one somewhere and sending me the link in a PM will also work.

[Stremon]

Originally Posted by davidfor

I could see the database being corrupted if the bad files caused a crash and reboot while the database was being written, but in the cases where the files are not processed, I don't think so. The firmware seems to process all the files and the database updates happen at the end in one go.

As to the crash, PeterT's suggestion is good. Look for some stack logs and supply a file that causes the problem. Kobo are the one who need to see it, but I'm curious to look as well. Ideally supplying a link to a freely available manga would be the best. But, putting one somewhere and sending me the link in a PM will also work.

True, my database has been corrupted few times after crashes/reboot. I had to rebuild it to make it work again.
I recently did a factory reset and didn't had any crash or reboot so far. I will send the stack files when I encounter it again.

By the way, do you know a simple and fast way to make shelf for comics and manga from my pc?
For books I do it automatically from calibre, but calibre is pretty bad and very slow to manage my cbr/cbz files...

[PeterT]

Calibre should be no different managing CBR / CBZ files than it is managing ePub filed.

The only required change would be to ensure that calibre knew your device supports those formats

[Angemon]

Stremon, try looking for non-image files on your comics before copying them to your e-reader. Sometimes they come with, for example, text files, or even the thumbs.db (a system file used for storing the thumbnails of the image files), and that can screw things up.

I do this every time I want to copy a comic to my Aura HD (I switched the 4GB SD card on my HD for a 32GB so I usually process dozens of volumes at once):

1 - Unpack files
2 - Search for non-image files
3 - Use Photoshop to batch rotate landscape pics to portrait
4 - Use Renamer to batch rename all files (ex. 00001.jpg, 00002.jpg...05489.jpg)
5 - Repack everything and use Renamer to change the extension to CBR/CBZ.


I never had any problem with any comic since I started doing that (and I once used a bash script to download ALL Dilbert strips since the first to the current date, I ended up with a CBR around 850MB and had no problems reading it - other than it took longer to open).

[Stremon]

Originally Posted by Angemon

Stremon, try looking for non-image files on your comics before copying them to your e-reader. Sometimes they come with, for example, text files, or even the thumbs.db (a system file used for storing the thumbnails of the image files), and that can screw things up.

I do this every time I want to copy a comic to my Aura HD (I switched the 4GB SD card on my HD for a 32GB so I usually process dozens of volumes at once):

1 - Unpack files
2 - Search for non-image files
3 - Use Photoshop to batch rotate landscape pics to portrait
4 - Use Renamer to batch rename all files (ex. 00001.jpg, 00002.jpg...05489.jpg)
5 - Repack everything and use Renamer to change the extension to CBR/CBZ.


I never had any problem with any comic since I started doing that (and I once used a bash script to download ALL Dilbert strips since the first to the current date, I ended up with a CBR around 850MB and had no problems reading it - other than it took longer to open).

Thank you very much, indeed one of the problems was the hidden thumbs.db which was included in the compressed files, I understood that too while writting my previous message...
There was also other problems which made the device bootlop and/or unable to read the comic properly, I will write them down just in case:
- files or folders inside the compressed files which contain non standard characters (like french è or é).
- images that contains corrupted data (even if they open up correctly in windows).
- images too big (I mean really really big)
- corrupted compressed file.
- any non supported file inside the compressed files.

The biggest problems where clearly the thumbs.db and special characters. I think people using only english and having a mac/lunix shouldn't have so much trouble haha

Thanks for your help guys, now I can finally make my kobo read everything!
I will modify the first post to put the solution, and help people who might need those informations too.

[Stremon]

Originally Posted by PeterT

Calibre should be no different managing CBR / CBZ files than it is managing ePub filed.

The only required change would be to ensure that calibre knew your device supports those formats

Actually Calibre always try to unzip every cbz or zbr and files and try to deal with every jpg file inside instead of taking them as a whole volume, and also try to rename them badly even if there is no metadata inside.
I heard somewhere that it is possible to compress every collection of cbr/cbz inside on big CBC file with metadata in a text file, but even if the solution seems cool, for me it is not really convenient.

[davidfor]

Firstly, thanks for the report on what was going wrong. About the only one I am surprised at is the extra files one. Not ignoring the thumbnail and similar files isn't good.

Originally Posted by Stremon

Actually Calibre always try to unzip every cbz or zbr and files and try to deal with every jpg file inside instead of taking them as a whole volume, and also try to rename them badly even if there is no metadata inside.

I'm not sure what you mean. When I have added a CBZ or CBR to calibre, it added the complete file as a book. The title and author were taken from the file name. I know that there is an informal standard for metadata inside these files, but I don't know if calibre reads this.

Any file renaming is to rename the CBZ/CBR to fit into the library file structure. This is fixed and uses the title and author.

[dem1980fr]

Originally Posted by Stremon

Actually Calibre always try to unzip every cbz or zbr and files and try to deal with every jpg file inside instead of taking them as a whole volume, and also try to rename them badly even if there is no metadata inside.
I heard somewhere that it is possible to compress every collection of cbr/cbz inside on big CBC file with metadata in a text file, but even if the solution seems cool, for me it is not really convenient.

Hello

i confirm that add a cbz into the Kobo with accent make it loops.
It try to create a thumbnail for the Home screen but reboots.
When i add an accent in a faile in the cbz i have this problem.
example test.cbz :
\chemin avec avec comme é\001.jpg
\chemin avec avec comme é\002.jpg

This file makes my Kobo to reboot (so don't do it !) and i have to restore it because i can't have access to its drive from computer anymore because of the loop reboot.
Nickel has always this problem in 4.32 firmware.

[davidfor]

Originally Posted by dem1980fr

Hello

i confirm that add a cbz into the Kobo with accent make it loops.
It try to create a thumbnail for the Home screen but reboots.
When i add an accent in a faile in the cbz i have this problem.
example test.cbz :
\chemin avec avec comme é\001.jpg
\chemin avec avec comme é\002.jpg

This file makes my Kobo to reboot (so don't do it !) and i have to restore it because i can't have access to its drive from computer anymore because of the loop reboot.
Nickel has always this problem in 4.32 firmware.

Firstly, did you notice that this was a thread from 2015? Which means it was for a 3.x firmware version?

And it is well know that Kobo has issues with some characters in file names. It is not limited to CBR/CBZ but can affect other formats. This is one reason to sideload from calibre. It will sanitize the file names when sending them to the device. That saves a lot of problems. But, it does not do any with the internal file names or directories of the books. These can cause problems and would need to be fixed in other ways.