Insights in Dogwood
I'd like to discuss how much we're going to support Insights going forward.
I know that we have documentation on how to set up Insights. What I'd like to know is if we think this is sufficient "support" for the time being for the community, or if there's something we can do more/better. We're planning to cut Dogwood in late October for a late November release. Between now and then I'd like to figure out what the best way to support our community wrt Insights will be.
Notes from meeting:
Generally Insights/pipeline/api needs a versioning scheme (3 repos!)
perhaps makes sense to tag the repos at the time when we cut named release
Do we want to bring Braden's solution into the fold?
Some investment to bring this into the configuration, but not yet enough for an average site to just bring it up
Big documentation gap
how to take the 3 repos and do something with them
Licensing is all taken care of (code is already open source)
Branding
Steps
"Beta":
Insights not at a push-button state, but some people have gotten it running. Take Dogwood versions, here's the rough docs (developer docs) we have, try it out, no guarantee/support from us
Find & document the version(s) of Insights that work with the named release
Doc how you get from code to working thing
It works super easy if you're on AWS and have an expense account! Works on EMR
Better docs for home hadoop install
Push button automation
TODOs
Open Source team: try standing this up prior to dogwood, and help improve documentation!
Definitely some missing documentation about the minimum steps & settings that you need to do to get this stood up
lms oauth handshake a pain - probably not well documented, and will not work until this is set up (Gabe unsure if any documentation exists at all)
Braden's new docs may be the best starting point
Validate branding (default instance should say Open edX, not use edX logos)
Open Source + Analytics team: consolidate documentation!
Victor: Make a wiki page to start collecting documentation
Open Source (@Joel Barciauskas (Deactivated)): What is the responsibility of the Analytics team in the process of getting the repos tagged, tested, etc?
Let's chat. We're working on an analytics devstack, based on Braden's initial work, so hopefully we'll have at least that much. The open question is whether we'll also have automation for production deployments. That's not scheduled at the moment.