"Unknown erorr" and "Server error" status syncing libraries since Seafile Server CE 7.0.5?


the last days I recognized that some libraries are out of sync even though the client indicated a synchronized state (green cloud symbol). Resyncing the libraries ended up in “Unknown error” or “Server error” library status in the client.

First thing I tried was purging the client and installing it again but it doesn’t help (same errors/status).

The last things I did on server side was upgrading from Seafile Server CE 6.3.4 to seafile-server_7.0.5_x86-64.tar.gz and a seaf-fsck.sh (with just one damaged video file listed) followed by a seaf-gc.sh (which cleared around 1TB data - it was my first garbage collecting running the server since several years now).

Suspicious server logs:

  • seafdav.log lists a lot of Failed to init seahub db: No module named seahub_settings.
  • controller.log lists a lot of pid file /opt/seafile/pids/seafdav.pid does not exist

(but I think is related to https://forum.seafile.com/t/seafile-7-0-2-seafdav-bug-no-module-named-seahub-settings/ or Seafdav does not work (seafdav.pid problem) ?)

  • seaf-fuse.log lists errors like
getattr.c(96): Path videos/education/documental/anthropology/dawn of mankind.smi doesn't exist in repo 1f4cf513-af6a-4e06-8ae2-13a85f7972f4.

for almost any file on my server. The file itself (dawn of mankind in this case) exists and I can access it on the server (but not as “smi” but as “flv” in this case).

All other server logs doesn’t contain errors.

Suspicious client logs:

  • applet.log: nothing I would call an error except some No such interface “org.freedesktop.DBus.Properties” on object warnings (maybe this is caused by the fact I’m using i3 as window manager under Debian)
  • seafile.log: Contains a lot of errors fo, example:
[01/04/20 02:10:29] sync-mgr.c(559): Repo 'ExampleRepo' sync state transition from 'synchronized' to 'uploading'.
[01/04/20 02:10:29] http-tx-mgr.c(1181): Transfer repo 'ec618343': ('normal', 'init') --> ('normal', 'check')
[01/04/20 02:10:29] http-tx-mgr.c(1181): Transfer repo 'ec618343': ('normal', 'check') --> ('normal', 'commit')
[01/04/20 02:10:29] http-tx-mgr.c(2767): Bad response code for PUT https://my.ddns.net/seafhttp/repo/ec617343-3123-46bd-8fd2-54ca70412e4a/commit/50dfcd1f69f4b48d163d2d43e1feac112028563b: 413.
[01/04/20 02:10:29] http-tx-mgr.c(3724): Failed to send head commit for repo ec618343.
[01/04/20 02:10:29] http-tx-mgr.c(1181): Transfer repo 'ec618343': ('normal', 'commit') --> ('error', 'finished')
[01/04/20 02:10:29] sync-mgr.c(621): Repo 'ExampleRepo' sync state transition from uploading to 'error': 'Error occurred in upload.'.

Any ideas?

Sorry, my fault. I’m using a separated reverse proxy and the value for client_max_body_size in the Nginx config of the reverse proxy was different from the value in the Nginx config of the Seafile server for some reason. If both are set to client_max_body_size 0; syncing works fine again.