I am using ext4 on a large external drive and I found an odd file with a bad description on the volume. So I decided to unmount it and run fsck over the drive. It wouldn't unmount though, says (quite classically) that is busy. I ran lsof over it and the pid associated with it was a journal instance which just did not want to let go.
I finally killed it and umounted the drive, but my feeling is that this shouldn't happen. The process was jbd2
I am wondering if there was a better way to handle this. I couldn't find a way of forcing jbd2 down
Don't use ext4: It's still 32-bit like ntfs and has other disadvantages.
I prefer xfs, when compression is not necessary, otherwise btrfs+zstd:2 (higher than btrfs:2 don't compress better and is a waste oft cpu time)
I appreciate your insight but this problem is with an ext4 file system 12 tetrabytes large. xfs is not ideal for it and I will not use btrfs. I don't like it. Regardless, comparative file systems is for a different thread. It might be worth opening a comparative file system thread to discuss those technologies in more detail, but it doesn't help with this problem.
I didn't want to compare. I had bad experiences with ext4, so I don't use it (since 10 years) and have no problems with it: That was my recommendation.
xfs seems to me nearly unbreakable, so I use it for important data and my backup drives. - And it shall be very good for large files: https://en.m.wikipedia.org/wiki/XFS
btrfs is good for compression, but easier to damage *), so I use it for the system partition only, because it is fast to restore.
*) the last time I had uncorrectable errors is 1,5 years ago, so it seems to become better.