cajal / pipecontrol Goto Github PK

Using either crontab or a custom python notification daemon will make duplicate requests. When the daemon only checks every five minutes, duplicate notifications appear less than five minutes later. When all crontab entries are deleted in the docker container, notifications are still sent out.

This happens when the datajoint connection gets closed after some time of inactivity (72 hours). I think we already tried reconnecting with the datajoint server (so run dj.conn() before any server call) but it didn't work (not sure what where the effects). Datajoint does have the dj.config['database.reconnect'] boolean flag so maybe this will solve it.

If it doesn't, we could add a daemon that asks for something to the flask server every 12 hours or so. This could even run inside the docker container (started in the ./run_server.sh script) so it's all self contained. And it should send some signal to the server just so it calls dj.ping() and keeps the connection alive.
So in ./run_server.sh

watch -n 43200 curl shikigami.ad.bcm.edu/keepalive 
# this won't quite work, we should daemonize it and put it before the `flask run` call so it
# returns control to the main thread and then `flask run` is executed, `flask run` should
# be the last thing to execute so we can access the flask server outputs via docker logs

and in views.py

@main.route('/keepalive')
def keepalive():
    dj.ping()
    return '', http.HTTPStatus.NO_CONTENT # https://www.erol.si/2018/03/flask-return-204-no-content-response/