From dc680b7db5d8b026650c4607659820778fedcc85 Mon Sep 17 00:00:00 2001
From: Stephen Greene <sgreene@redhat.com>
Date: Thu, 6 Aug 2020 10:15:04 -0400
Subject: [PATCH 1/2] Add new router failed reload counter metric

If a router reload fails after a router pod becomes
ready, we need a way to alert cluster admins that
newly created route resources are not being applied
to the cluster.
---
 pkg/router/template/router.go | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/pkg/router/template/router.go b/pkg/router/template/router.go
index 0ccd22fd1..85f9539ab 100644
--- a/pkg/router/template/router.go
+++ b/pkg/router/template/router.go
@@ -98,6 +98,8 @@ type templateRouter struct {
 	stateChanged bool
 	// metricReload tracks reloads
 	metricReload prometheus.Summary
+	// metricReloadFails tracks reload failures
+	metricReloadFails prometheus.Counter
 	// metricWriteConfig tracks writing config
 	metricWriteConfig prometheus.Summary
 	// dynamicConfigManager configures route changes dynamically on the
@@ -178,6 +180,12 @@ func newTemplateRouter(cfg templateRouterCfg) (*templateRouter, error) {
 		Help:      "Measures the time spent reloading the router in seconds.",
 	})
 	prometheus.MustRegister(metricsReload)
+	metricReloadFails := prometheus.NewCounter(prometheus.CounterOpts{
+		Namespace: "template_router",
+		Name:      "reload_fails",
+		Help:      "Tracks the number of failed router reloads",
+	})
+	prometheus.MustRegister(metricReloadFails)
 	metricWriteConfig := prometheus.NewSummary(prometheus.SummaryOpts{
 		Namespace: "template_router",
 		Name:      "write_config_seconds",
@@ -207,6 +215,7 @@ func newTemplateRouter(cfg templateRouterCfg) (*templateRouter, error) {
 		dynamicConfigManager:     cfg.dynamicConfigManager,
 
 		metricReload:      metricsReload,
+		metricReloadFails: metricReloadFails,
 		metricWriteConfig: metricWriteConfig,
 
 		rateLimitedCommitFunction: nil,
@@ -434,6 +443,8 @@ func (r *templateRouter) commitAndReload() error {
 		if r.dynamicConfigManager != nil {
 			r.dynamicConfigManager.Notify(RouterEventReloadError)
 		}
+		// Increment the failed reload counter when a reload fails
+		r.metricReloadFails.Inc()
 		return err
 	}
 

From ad0336b1d1c300aabd2b6941b7f39998a4147a24 Mon Sep 17 00:00:00 2001
From: Stephen Greene <sgreene@redhat.com>
Date: Thu, 6 Aug 2020 10:15:24 -0400
Subject: [PATCH 2/2] Remove premature router reload call at startup

The first call to `commitAndReload` bypasses the rate limited
reload logic that also takes into account route sync status.
Removing this initial router reload call will prevent the router
from starting in a "routeless" state: that is, a state were
the router is running before it has begun watching route
resources.
---
 pkg/router/template/router.go | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/pkg/router/template/router.go b/pkg/router/template/router.go
index 85f9539ab..a65a57f21 100644
--- a/pkg/router/template/router.go
+++ b/pkg/router/template/router.go
@@ -230,10 +230,7 @@ func newTemplateRouter(cfg templateRouterCfg) (*templateRouter, error) {
 		log.V(0).Info("initializing dynamic config manager ... ")
 		router.dynamicConfigManager.Initialize(router, router.defaultCertificatePath)
 	}
-	log.V(4).Info("committing state")
-	// Bypass the rate limiter to ensure the first sync will be
-	// committed without delay.
-	router.commitAndReload()
+
 	return router, nil
 }