Problems during backup using seaf-fuse

Seafile Server CE version 6.0.9 running on Debian as user seafile.
As user seafile run start /var/seaf-fuse
Everything ok, I can access files from every library, e.g. /var/seaf-fuse/ Biblioteca
Now I start a remote backup from a backup host running backuppc which in turn runs a remote rsync command on Seafile Server:
/usr/bin/rsync --server --sender --numeric-ids --perms --owner --group -D --links --hard-links --times --block-size=2048 --recursive . /var/seaf-fuse
This is the standard command for taking backups remotely with backuppc.
The backup ends prematurely, just a few files and folders are transferred,
rsync errors state that files and directories have vanished, and in seaf-fuse.log I get lots (6500 lines) of errors such as:

04/27/17 17:17:04] …/common/seaf-db.c(179): Error exec query SELECT repo_id FROM Repo WHERE repo_id = ‘79e901eb-a5d5-4f5e-a91f-59e24654cf41’: sqlite3_step failed: database is locked.
[04/27/17 17:17:04] seaf-fuse.c(197): Failed to get repo 79e901eb-a5d5-4f5e-a91f-59e24654cf41.
at a rate of 400 lines/sec aprox.

This server has 150 libraries and 40 users aprox.
I read Exposing seaf-fuse to another server, any idea?


For such a number I highly recommend using MySQL.

I read Is it possible to undo this migration?
I think its not so easy once the new mysql tables are updated. It would be nice to have a switch in seafile/seahub to start the entire system in read-only mode, so that no tables or seafile-data is updated. In this mode the old sqlite tables could be restored if something fails during or after migration.