Provide status information for when pod deployments fail

[jiekang]

Parent: #1658

There needs to be some way for the UI to know when a pod deployment fails.

In Openshift, there is a websocket connection to, for example, wss://api.starter-us-east-2.openshift.com/api/v1/namespaces/jkang-run/events?watch=true&resourceVersion=928801073 that provides information like:

"Error creating: pods "osio-prod-1-1-" is forbidden: exceeded quota: compute-resources, requested: limits.cpu=1,limits.memory=512Mi, used: limits.cpu=2,limits.memory=1Gi, limited: limits.cpu=2,limits.memory=1Gi"

Since we don't have websockets at this time, we either need an alternative, or to move forward with websocket support.

Alternatives:

1. Add an API that the UI polls for events.

2. UI calculates based on quota information when attempts to scale are made and rejects ones that would go over the quota.

[jiekang]

@ebaron @stooke Would you be able to look at what it would take to provide an API for this?

[jiekang]

Note option 2 does not cover the situation when users create new deployments that exceed quota externally from the page, for example by importing an existing application or using the quickstart wizard. Looks like some form of option 1 would be best. Even then, I'm not sure the events API would cover that scenario unless we looked at all past events.

[jiekang]

Related to (Deployments): Use websockets instead of HTTP requests #2380

Events API is best served through websockets.

[jiekang]

Closing as this is tracked via

#874
#2380
#3344
#3346