Running your own server

Plom has several components. One of these is the Plom server, used both for preparing new assessments and for coordinating grading.

This page details running a publicly exposed server for production use with live data. If you are interested in running a server for non-production use, please see Installing the Plom Server.

Note

We’re not experts on hosting services. The following are some suggestions and things Plom server admins should keep in mind, based on our experience. Your mileage may vary!

Containerisation

Each Plom container requires its own database. Currently these are limited to PostgreSQL. One can either run one PostgreSQL server container per Plom container, or share a PostgreSQL server between multiple Plom containers. Each Plom server release is documented on gitlab, and container images are available and currently hosted on DockerHub.

Security

While the Plom container contains some security measures, it would be unsafe (and irresponsible) to expose the minimally viable plom``+``postgres containers on a public network without a reverse-proxy. Nginx, and Apache are both good options for a reverse proxy.

Backups

Make them. An idealised Plom container is stateless, but the current iteration unfortunately carries some state: you should backup the container volume of the Plom container and appropriate backups of the database.

OrcaSequestration

The Plom project provides some scripts for deploying and managing production servers, contained in the OrcaSequestration. repository.

These orca scripts are designed to be easy to use and broadly applicable, at the cost of some efficiency (e.g., it’s not necessary for each plom container to have a separate nginx reverse-proxy, but this simplifies deployment for the admin). This approach is subject to change and patches are welcome.