Seafile client does not sync 1 file in Windows 10 and says 'synced'

Hi!

I just added around 50 GB to one library that is synced between two computers. Afterwords I compared the size of the directories: On the other client computer some 200 MB are missing. So I compared the content (using FreeFileSync) and found mainly ~XXXX.tmp files missng - I think I can live with that.
But one file is just not uploaded:
It is this file:
C:/Users/ildiko/Documents/minipc sync seafile/Seafile/Meine Bibliothek/Aufnahmen/Vergiss nie_ was du gesehen hast__ Krimi von Ilkka Remes (Ilkka Remes) - 2013 - Alle Hoerspiele-00.mp3
The file (also) does not show up in the webbrowser view of the library.
Renamed the file in order to make seafile recognize the change. Seafile in fact starts doing something, but the file is still not uploaded.
I openend the seafile.log and found:

[03/31/18 10:01:00] http-tx-mgr.c(2325): Bad response code for POST http://ubu.fritz.box:8082/repo/head-commits-multi/: 500
[03/31/18 10:01:30] http-tx-mgr.c(2325): Bad response code for POST http://ubu.fritz.box:8082/repo/head-commits-multi/: 500
[03/31/18 10:02:00] http-tx-mgr.c(2325): Bad response code for POST http://ubu.fritz.box:8082/repo/head-commits-multi/: 500
[03/31/18 10:02:24] sync-mgr.c(739): Repo 'Meine Bibliothek' sync state transition from 'synchronized' to 'committing'.
[03/31/18 10:02:30] http-tx-mgr.c(2325): Bad response code for POST http://ubu.fritz.box:8082/repo/head-commits-multi/: 500
[03/31/18 10:02:36] cdc.c(160): CDC: failed to read: Permission denied.
[03/31/18 10:02:36] ../common/fs-mgr.c(697): Failed to chunk file with CDC.
[03/31/18 10:02:36] repo-mgr.c(1212): Failed to index file C:/Users/ildiko/Documents/minipc sync seafile/Seafile/Meine Bibliothek/Aufnahmen/Vergiss nie_ was du gesehen hast__ Krimi von Ilkka Remes (Ilkka Remes) - 2013 - Alle Hoerspiele-00.mp3.
[03/31/18 10:02:36] repo-mgr.c(3741): All events are processed for repo ad3f6424-22c1-4939-b7d6-057356da4135.
[03/31/18 10:02:36] sync-mgr.c(739): Repo 'Meine Bibliothek' sync state transition from 'committing' to 'synchronized'.
[03/31/18 10:02:55] sync-mgr.c(1553): Removing blocks for repo Meine Bibliothek(ad3f6424).
[03/31/18 10:03:00] http-tx-mgr.c(2325): Bad response code for POST http://ubu.fritz.box:8082/repo/head-commits-multi/: 500
[03/31/18 10:03:30] http-tx-mgr.c(2325): Bad response code for POST http://ubu.fritz.box:8082/repo/head-commits-multi/: 500
[03/31/18 10:03:50] sync-mgr.c(739): Repo 'Meine Bibliothek' sync state transition from 'synchronized' to 'committing'.
[03/31/18 10:04:00] http-tx-mgr.c(2325): Bad response code for POST http://ubu.fritz.box:8082/repo/head-commits-multi/: 500
[03/31/18 10:04:02] cdc.c(160): CDC: failed to read: Permission denied.
[03/31/18 10:04:02] ../common/fs-mgr.c(697): Failed to chunk file with CDC.
[03/31/18 10:04:02] repo-mgr.c(1212): Failed to index file C:/Users/ildiko/Documents/minipc sync seafile/Seafile/Meine Bibliothek/Aufnahmen/Vergiss nie_ was du gesehen hast__ Krimi von Ilkka Remes (Ilkka Remes) - 2013 - Alle Hoerspiele_00.mp3.
[03/31/18 10:04:02] repo-mgr.c(3741): All events are processed for repo ad3f6424-22c1-4939-b7d6-057356da4135.
[03/31/18 10:04:02] sync-mgr.c(739): Repo 'Meine Bibliothek' sync state transition from 'committing' to 'synchronized'.
[03/31/18 10:04:21] sync-mgr.c(1553): Removing blocks for repo Meine Bibliothek(ad3f6424).
[03/31/18 10:04:30] http-tx-mgr.c(2325): Bad response code for POST http://ubu.fritz.box:8082/repo/head-commits-multi/: 500

There is a message saying “read: Permission denied” in the time context of that file. So I thought, something wrong with it. But I could easily open it and listen to the radio drama.
BTW: No other programs are running on the machine (that might have a file lock).

  1. What might be the reason for this problem?

  2. should I care about the masseges “Bad response code…”?

  3. Another thing: I absolutely do not like that seafile client indicating “synced” in spite of the fact that one file is not uploaded.

Best regards,
Pfeffer.

The problem is not a Seafile problem. It’s a file system problem with Linux. You have a “(” and a “)” in the filename. Those cannot be used in filenames in a Linux environment.

Another issue is the path is very long. I haven’t counted the number of characters, but if it exceeds the path size, the file will be “stuck”. For whatever reason, Windows allows files to be copied to a hard drive with long file names and doesn’t compute the total number of characters the path would have. Most likely, though, even if the ( and ) weren’t in the filename, if the path size would be too long inside Linux, it would fail to sync as well.

Back when I collected MP3 files, I very often had that problem with files getting stuck. You can’t even delete them by conventional means when it happens.

So, the first thing I would try is changing that filename of that mp3 and see what happens.

[EDIT] - I think I may have confused the ( and ). They may be able to be used in filenames, but handling of them inside a script in Linux usually requires enclosing them with ’ and '. I’ll do further research, but I’m betting if you shortened that filename or path that it’s in, it will sync fine.

Ok… .I just confirmed that Seafile will sync files with a ( or ) in them. I’m betting your problem is the very long path.

By the way, the Bad response error is returning error 500, which is an internal server error on http://ubu.fritz.box:8082/repo/head-commits-multi/. Whether or not to be concerned with it, I’m not certain yet. I’m not familiar with “head-commits-multi” folder.

Ok, I tracked it down in the code. That Bad response does not appear related to your problem. Note that the Bad Response occurs exactly every 30 seconds, and it’s happening on a folder /repo/head-commits-multi, which does not exist in my Seafile folder structure.

In short, I don’t know that it’s anything to worry about. Maybe someone else can chime in on that.

As for your main problem, Seafile is unable to index that particular file. Try making a copy of it in that folder and name it to test.mp3. See if it then syncs. If it does, then I’m suspecting that the resulting path is just too lengthy.