[RFC] Installation Health Check for Sunbird

[parthlawate]

PROBLEM STATEMENT
I've seen a lot of early adopters struggle with installation of Sunbird & new issues with installation keep cropping up as more aspects are tested. This can help onboard newer developers & adopters quicker.

PROPOSED SOLUTION
It would be nice to as a first step create a 'Installation Health Check' script that can check aspects that are missing, what services are not up , what aspects are not configured etc & show their status. We can also point to relevant documentation where some aspect is 'red'.

Would love thoughts & comments on this. cc @coolbung @vrayulu

[coolbung]

Some thoughts on which health checks would be useful

Services

• Check services using server side checks (i.e. player being able to communicate to all services health endpoints)
• Check API endpoints with client side checks to identify any CORS etc issues
• Check external connections like Azure blob

Configuration checks

• Check and report key configuration variables are set
• Check if other specific things video streaming settings are in place

For background jobs, it would be good to have a health.ping kafka topic where a health check event is added, and all jobs read events from this topic, perform their internal health checks and respond on a health.pong topic. The portal can then read from this topic and show the health of the backend jobs.

[pramodkvarma]

Excellent idea! Btw, we should NOT look at Sunbird as a monolith, rather, think of them as building blocks and have each building block contributor build "installation health check".

This is super critical. Otherwise, we will end up building some all inclusive health check tool. @vrayulu @madhucr

[parthlawate]

So in effect we should do health check scripts on Building Blocks level & a 'collection' view can be built for Applications that combine different building blocks.. Maybe we can do a pilot one in sunbird RC ?

[pramodkvarma]

Yes! Agree, we should walk the talk by implementing in 2 BBs (one is a never a good sample). Take Sunbird-RC (pure building block) and maybe cQube (a solution). Then we should do one for Sunbird-Ed (large solution) once we have a good template in place.

[parthlawate]

Yes multiple is a better .. Any other ones @coolbung from our experience from the ones being especially tricky which will give the adopters some quick benefit ?