Andrew Dryga

Two questions/design tradeoff decisions: 1. Do we show resources that the client can't access right now due to a condition violation? (Eg. "not from this country")? Yes, because having the...

Deleting entities is a rare action so I think it's better to leave it where it is now. Otherwise, we will have a big red button that always distracts you...

@bmanifold yeah, converted to draft, ty

@jamilbk I think the portal doesn't need to do that selection, the client already has all the filtering rules so it can decide which resource to use locally

We are missing the docs intent here. It's not that they recommend replacing Cloud Metrics with Prometheus, they just deprecated a direct way to export StackDriver metrics in Prometheus format...

I might be wrong but I did not find any docs about managed Prometheus for non Kubernetes environments, eg. https://cloud.google.com/stackdriver/docs/managed-prometheus/setup-managed. This means we will need to run everything in Kubernetes...

Would building a client for https://cloud.google.com/monitoring/api/metrics_other require a lot of work? I assume it's just a buffer data structure (eg. ring buffer) with overflow handling + client lib with a...

We might use the same metrics for #4415.

@jamilbk do we even get blocked traffic at the gateway nowadays or it should be filtered on the client already? cc @conectado

On the portal end two processes are responsible for communicating with a WebSocket (this is an oversimplification but the rest are not important to you): - socket process - responsible...