Dashboard produces excessive amount of requests

[sclu1034]

Summary

According to #2776, keeping the dashboard open used to trigger two requests every 5 seconds.
Apparently, that has doubled since then, as I'm now seeing four (resources, jobs, teams, pipelines). That's almost 3000 requests per hour, where the overwhelming majority contain the exact same data as their previous call.

Steps to reproduce

1. Open the dashboard of any Concourse instance
2. Open the browser's Developer Tools network tab
3. Wait a few seconds

Expected results

Short term, I would have expected these to be combined into a single "get dashboard data" call.

Long term, I would expect this to be implemented with long polling or some form of server-side events (e.g. websockets), rather than a continuous stream of requests.

And in either case, the responses seem to always return the full data set. I would expect that after the initial load of the dashboard, subsequent requests only return what's changed since the previous one (kind of a requisite for the polling/SSE anyways).

Actual results

Currently, there are several separate requests fired at the same time, that always return the full data set despite it being identical to the previous hundreds of requests.

Additional context

N/A

Triaging info

• Concourse version: v7.11.0
• Browser (if applicable): Firefox
• Did this used to work?

[taylorsilva]

Just want to say that you are 100% correct about the current behaviour and for laying out what some improvements this could look like. If someone wants to help out with this... 😬