PROJECT HAS MOVED

Note that the project has moved to the Gitlab Xen Project space, here:

https://gitlab.com/xen-project/misc/session-scheduler

Intro

session-scheduler is a tool for scheduling Unconference-style discussion sections to maximize participation. The basic idea is:

1. The event creators define the "slots" available for sessions
2. Members suggest sessions
3. Attendees express how interested they are in attending various sessions
4. The scheduler tries to find a schedule that maximizes utility

This is similar to Ubuntu's Summit scheduler, but:

1. Lighter-weight; doesn't require a launchpad account, anybody can create sessions, &c
2. Allows users to express different levels of interest

At the moment, this is a prototype focused on Xen's Developer Summit.

Build instructions

Clone the repo:

git clone https://github.com/gwd/session-scheduler
cd session-scheduler

Build the binary:

go build

Starting instructions

Running it is simple. From the git repo, run:

./session-scheduler

The first time you run session-scheduler it will create a number of databases in data/:

• serverconfig.sqlite to store data about configuration: default port number, admin password, and so on, so that you don't have to enter these in every time
• event.sqlite to store data about the event itself: timetables, locations, users, discussions, interest
• sessions.sqlite to store data about active user sessions. (This means you can restart your server and logged-in users are still logged in.)

It will also create an account named admin, generate a random password, and print the password to stdout; and if the port number hasn't been specified, it will generate a random port number and bind to it.

./session-scheduler
2020/05/11 15:34:13 Default location to Europe/Berlin
2020/05/11 15:34:13 New user post: 'admin'
2020/05/11 15:34:13 Administrator account: admin DB7Zixb2RYra
2020/05/11 15:34:13 Listening on localhost:22752

Admin mode

Logging in as admin, you are in "Admin mode". The admin account may edit users and sessions, but not express interest in sessions. If you want to experience life as a user, you'll have to create a user account.

The admin account has a "Console" page available. From there you can initiate the session scheduler and enable test mode, set the verification code, and other admin activities.

Deployment

To run elsewhere without cloning the entire repo, copy the sessions-scheduler binary, along with the following directories: assets templates. Also create a directory, data, for session-scheduler to store the databases.

deploy.sh has been created to do this automatically; it's hard-coded with our server, user, and directory, so modify it according to your use case.

There is a -serverlock method you can use to handle potential duplicate copies of the same server; quit will exit if another instance is running, error will exit with an error, and wait will wait for the other instance to quit.

Current recommended practice is to start the server in a tmux session, like so:

./session-scheduler -servelock quit >> sessions-scheduler.log &
tail -f sessions-scheduler.log

And then add a crontab like the following:

@reboot /home/xensched/session-scheduler/session-scheduler -servelock quit >> /home/xensched/session-scheduler/session-scheduler.log

Backups

The most robust way to create automatic backups is to use the sqlite3 tool with the VACUUM command, and put it in your crontab. Below is an example crontab entry:

0 4 * * * sqlite3 session-scheduler/data/event.sqlite "vacuum into \"backup/event-$(date -Iminutes).sqlite\""

This is also handy for getting a copy of the "live" database to do debugging.

Miscellaneous notes

Registration requires a verification code. This will be generated randomly the first time session-scheduler is run. It can also be set from the admin console.

Discussion proposals refuse discussions with duplicate titles.

Any given user is limited to generating sessions equal to the total number of session slots.