Poll for assemblage updates in the downstream

The ProxyAssemblage controller does not have the opportunity to be
notified of changes in a downstream cluster (or at least, this would
be tricky to arrange, especially at scale), so it won't in general see
when its Assemblage counterpart changes status.

A dumb but effective way to make sure updates are seen is to simply
requeue each ProxyAssemblage to be re-examined after a small
interval. This will be OK at moderate scale (I would guess O(100)
clusters), since more worker threads can be configured to cope with
polling. Larger scales may need a less brute force approach to be
figured out.

Signed-off-by: Michael Bridgen <mikeb@squaremobius.net>

Author: squaremo

module/controllers/proxyassemblage_controller.go

@@ -8,6 +8,7 @@ import (
 	"context"
 	"fmt"
 	"strings"
+	"time"
 
 	"k8s.io/apimachinery/pkg/runtime"
 	"sigs.k8s.io/cluster-api/controllers/remote"
@@ -20,6 +21,8 @@ import (
 	fleetv1 "github.com/squaremo/fleeet/module/api/v1alpha1"
 )
 
+const proxyPollInterval = 20 * time.Second
+
 // ProxyAssemblageReconciler reconciles a ProxyAssemblage object
 type ProxyAssemblageReconciler struct {
 	client.Client
@@ -84,7 +87,8 @@ func (r *ProxyAssemblageReconciler) Reconcile(ctx context.Context, req ctrl.Requ
 		return ctrl.Result{}, err
 	}
 
-	return ctrl.Result{}, nil
+	// We don't get notified of things happening in the downstream cluster; so, requeue to poll instead.
+	return ctrl.Result{RequeueAfter: proxyPollInterval}, nil
 }
 
 // SetupWithManager sets up the controller with the Manager.