Staging server price ball-park

I’ve been wanting us to have a staging server for our Discourse and Graphryder setup for some time, but the issue is unfortunately that we don’t have the budget or manpower to set up or maintain a staging server.

There are things that I or @owen or @gdpelican could do to offload @daniel and @matthias, but it’s hard unless there is a staging server where making mistakes doesn’t lead to production downtime. I think it’s important to have that sort of onboarding experience for new IT-related staff as we scale.

If we had a staging server, we could then install new versions of the Graphryder plugin ourselves and run tests with the same point of reference. For example, if the staging server was loaded with dummy test data (threads, codes, and annotations) for SSNA development, we could try out new plugins and features with ethnographers and researchers without having to apply them to real data.

It’s important for future budgeting that we know what the ballpark cost would be.

@matthias and @daniel – what do you think the ballpark cost is to:


  • Set up an additional server with our fork of Discourse
  • Document the set-up instructions so that they can be understood and replicated
  • Document some important routine maintenance, like backups, upgrades, and installing and upgrading new plugins


  • Monthly cost
  • Added overhead in terms of EUR for time spent on upgrading the staging server along with production (this can also be done by someone else than @matthias or @daniel as long as we have instructions)

I will then work towards the one-off cost being included in a project budget, and for the maintenance to be covered by overhead.

A staging server would be useful to get a better idea of how Discourse works (I know I can set it up on my own machine - but having an online sandbox would make things a lot easier)… but also for prototyping potential ideas.

An example: we may (or may not, depending on the demand) want a simple extension to easily insert webkit tags into a topic. This would make building sites a lot easier, and reduce the likelihood of mistyped tags. The extension might also render a sort of thumbnail in the post to indicate it is a webkit component - I’m not sure. The point is…having an environment available would be an advantage for any developer experimenting with such concepts.

just my 2¢

Not sure how powerful a machine you need, but you can get a t3.micro instance from Amazon for one year for free. It’s not very expensive after that.
Also, it sounds like some kind of deployment strategy would be useful (Ansible?), since in addition to setting up new staging/production servers it would be useful for people interested in running a local version on their workstation.
In terms of dummy data, I wonder if something like factory_boy which generates test data would be considered. In addition to generating test user data, you can use to create data for regression tests, which can make the program more robust.