Having used MB Picard on Windows for years without this issue, I am suprised that tagging on my new Linux PC is so slow; tagging one album (flac 500MB) takes 25 minutes.
Where should I look for a possible solution?
I attach debug log:
It is a known issue that saving large files to a SAMBA share is horrible slow, especially so if the files are large and have to be fully rewritten. If the files don’t need to be resized it is usually reasonable fast, even over network shares.
The best workaround is to tag the files on your local hard drive, then move it back to the share.
So far I think we don’t really now what is to blame here. My personal suspicion is that the way the files are saved very efficiently on local drives ¹) probably does not work on the network share. And on saving the file it constantly tries to use the faster code, gets an error and falls back to slower code. And this all costs time, resulting in these extremely long durations.
See also the following ticket:
¹) For the technical details: saving locally uses memory mapped files, which allows for very fast manipulation of file contents. But this probably does not work on mounted network shares.
I save everything locally, then use rsync to send everything over samba. While connecting to an rsync daemon or ssh (how I connect to one of the storage spaces) is much faster and far more efficient, I use the same tool with a very similar set of options to sync to a router with a USB drive connected (WIndows NT running on that to some capacity).
If your destination for the files is another linux machine, I would recommend transferring over ssh or setting up the rsynd (it’s much easier and faster to setup using ssh in my opinion) and copy the files outside of picard. Even if the transfer must happen over samba, you’ll see an increase in performance by cutting picard out of that process, and even more so if done via the terminal rather than a GUI file manager. If you’re not comfortable with the terminal, I believe grsync should be available in most repos and gives a UI to assist the process.