move eventsources out of proc to containers

[vaikas]

Fixes Issue #

Proposed Changes

• Move Event Sources from running within the controller to their own pods. Make them more 'distribution friendly' so they are self contained and easier to develop and can be done outside the controller.
• Introduce an event source runner and show how the GCP pubsub handles the long running operations. Same approach should also work for example like legacy source (database / filesystem notifier for example).
• Remove the receive adapter image hack and introduce parameters to Event Sources.
• Introduce an event_source_wrapper that makes it easier to write event sources as it fetches the necessary parameters.

[google-prow-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: vaikas-google

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• OWNERS [vaikas-google]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[grantr]

/lgtm

TIL about termination message!

[grantr]

Why doesn't this return err? Should we retry? Seems like it should be Error log level at least. Why was the BindFailed update removed?

[grantr]

WDYT about making this a defer?

defer func() {
  if err := t.delete(pod); err != nil {
    glog.Infof("failed to delete the pod after running: %s", err)
  }
}()
if _, err = t.run(pod, false); err != nil {
  glog.Infof("failed to unbind pod: %s", err)
}
return err

Same applies to Bind

[grantr]

https://godoc.org/k8s.io/apimachinery/pkg/util/wait#PollUntil might be useful here