So I need to build a WebSocket API for my org. The requirements from the business are pretty typical websocket pattern stuff except for one detail:
This websocket api will be used by different teams in our org, and each team needs its own separate activeconnections dynamodb table.
Now in a typical websocket api, there would be a single connections table that the connect and disconnect lambda functions write/delete to. Also, the hooks in the websocket api ensure that the connectionId needed to identify a connection/session are always in the event.requestContext. Easy peasy for a single connections table.
However, In my approach of having a separate active connections db/table for each team, it gets more complicated. Yes, it's true that for the connect lambda, It is very easy to code so that it expects a "TeamDatabaseID" from somewhere in the initial connection request - Headers, queryStringParameters, etc.
The problem is in the subsequent disconnect that could be triggered from either client or server. The disconnect hook will run the disconnect function, and pass in the default requestContext with the connectionId, but with no TeamDatabaseID - which the disconnect lambda needs to have access to in order to know which database to delete from.
Is there a way to do this? Is there some notion of a context object that I can set values in from the initial connection, so that when the disconnect happens, the teamDatabaseID is propagated in some way to the subsequent disconnect lambda? I tried writing to the requestContext - and that seems to only be alive for the execution of the given lambda.
