Seafdav fails to start after CE upgrade to 7.1.3

I’ve upgraded my CE (on CentOS 7) to Seafile 7.1.3. Most things seems to be working, except the seafdav server. The controller tries in a loop to run it with

/opt/seafile/bin/python3 -m wsgidav.server.server_cli --server gunicorn --root / --log-file /opt/seafile/logs/seafdav.log --pid /opt/seafile/pids/ --port 8080 --host

But it fails to start. When running this command manually as the seafile user, here’s the error :

[seafile@seafile seafile-server]$ /opt/seafile/bin/python3 -m wsgidav.server.server_cli --server gunicorn --root / --log-file /opt/seafile/logs/seafdav.log --pid /opt/seafile/pids/ --port 8080 --host
usage: wsgidav [-h] [-p PORT] [-H HOST] [-r ROOT_PATH]
               [--auth {anonymous,nt,pam-login}]
               [--server {paste,gevent,cheroot,cherrypy,ext-wsgiutils,flup-fcgi,flup-fcgi_fork,wsgiref}]
               [--ssl-adapter {builtin,pyopenssl}] [-v | -q]
               [-c CONFIG_FILE | --no-config] [-V]
wsgidav: error: argument --server: invalid choice: 'gunicorn' (choose from 'paste', 'gevent', 'cheroot', 'cherrypy', 'ext-wsgiutils', 'flup-fcgi', 'flup-fcgi_fork', 'wsgiref')
[seafile@seafile seafile-server]$

Looks like wsgidav does not support gunicorn. Anyone else has this issue ?

It works on my end.

Do you have gunicorn installed?

you did’t export the necessary variables before you did run the command.

OK, found my problem. I’m running seafile from a venv, and by mistake, I installed wsgidav (upstream) in the venv, which was used instead of the wsgidav version bundled with seafile. After removing wsgidav from the venv, everything is working