Hi, as much as we appreciate having a dedicated OCL space on Open Concept Lab - OpenMRS Talk we’ve started considering running a separate discourse instance for OCL in order to be able to support our community and have an OCL branded communication medium. We’d love your input on self-hosting vs managed instance and tap on your expertise with managing discourse for OpenMRS. We have some questions to be answered, but any opinion is welcome.
Given there’s space on jetstream, would it be much effort to deploy and maintain another instance specific to OCL?
How much effort is it to keep discourse up-to-date and behaving well? How often do we upgrade and how much time does it take given the size of the current OpenMRS instance?
How big is the current OpenMRS instance in terms of storage and views?
I understand some of these may not be public data or not easily to be answered so please share what you can. Would you recommend any other asynchronous communication channel that can be easily searched with google and is easy to follow? Thanks!
FWIW I think this direction makes sense. Although if it doesn’t make sense logistically then of course let’s continue the current course. What do active users like @wamz and @ball think?
@burke@Jen we can probably share a screenshot of the Talk metrics here, right?
Bit of an unnecessary soapbox, but I hope you will choose a discourse layout similar to OpenMRS… Personally I find it overwhelming when every single post is in different categories on other global good forums, because it feels like you have to click every folder to see what is going on.
We could certainly host an instance of Discourse in Jetstream, but I don’t think we could manage another instance of Discourse – i.e., it would be up to the OCL team to set up and maintain the instance.
I believe the technical setup is fairly straightforward (I wasn’t involved in the initial setup, but I think it’s pretty much a git clone and configure). Setting it up in Jetstream would involve making/deploying changes with our terraform & ansible repositories, setting up backups, etc. So, there’s probably a modest setup effort involved. Once set up, then there’s some additional modest effort in getting things configured (branding, categories, badges, etc.).
Discourse is one of our more stable services. I perform upgrades weekly almost always through the web UI with zero downtime (rarely an update requires SSH for a git pull and manual rebuild, which causes ~5-10 min of downtime). Most of the effort in “maintenance” is managing content (moderation of posts, approving false positive spam detection or posts from people trying to post immediately after joining, etc.).
Our instance is currently just under 30G with a few thousand visits and 200-300K views per month.