Merge pull request #13603 from kargakis/update-retry-count

Merged by openshift-bot

Author: OpenShift Bot

pkg/deploy/controller/deployment/controller.go

@@ -22,9 +22,10 @@ import (
 
 // maxRetryCount is the maximum number of times the controller will retry errors.
 // The first requeue is after 5ms and subsequent requeues grow exponentially.
-// This effectively can extend up to 5^10ms which caps to 1000s:
+// This effectively can extend up to 5*2^14ms which caps to 82s:
+//
+// 5ms, 10ms, 20ms, 40ms, 80ms, 160ms, 320ms, 640ms, 1.3s, 2.6s, 5.1s, 10.2s, 20.4s, 41s, 82s
 //
-// 5ms, 25ms, 125ms, 625ms, 3s, 16s, 78s, 390s, 1000s, 1000s
 //
 // The most common errors are:
 //
@@ -33,7 +34,7 @@ import (
 // * pod may be missing from the cache once the deployment transitions to Pending.
 //
 // In most cases, we shouldn't need to retry up to maxRetryCount...
-const maxRetryCount = 10
+const maxRetryCount = 15
 
 // fatalError is an error which can't be retried.
 type fatalError string

pkg/deploy/controller/deploymentconfig/controller.go

@@ -24,6 +24,12 @@ import (
 	deployutil "github.com/openshift/origin/pkg/deploy/util"
 )
 
+const (
+	// maxRetryCount is the number of times a deployment config will be retried before it is dropped out
+	// of the queue.
+	maxRetryCount = 15
+)
+
 // fatalError is an error which can't be retried.
 type fatalError string
 
@@ -389,13 +395,14 @@ func (c *DeploymentConfigController) handleErr(err error, key interface{}) {
 		return
 	}
 
-	if c.queue.NumRequeues(key) < MaxRetries {
+	if c.queue.NumRequeues(key) < maxRetryCount {
 		glog.V(2).Infof("Error syncing deployment config %v: %v", key, err)
 		c.queue.AddRateLimited(key)
 		return
 	}
 
 	utilruntime.HandleError(err)
+	glog.V(2).Infof("Dropping deployment config %q out of the queue: %v", key, err)
 	c.queue.Forget(key)
 }
 

pkg/deploy/controller/deploymentconfig/factory.go

@@ -24,9 +24,6 @@ const (
 	// controller stores have synced. If it hasn't synced, to avoid a hot loop, we'll wait this long
 	// between checks.
 	storeSyncedPollPeriod = 100 * time.Millisecond
-	// MaxRetries is the number of times a deployment config will be retried before it is dropped out
-	// of the queue.
-	MaxRetries = 5
 )
 
 // NewDeploymentConfigController creates a new DeploymentConfigController.

pkg/deploy/controller/generictrigger/controller.go

@@ -12,6 +12,12 @@ import (
 	deployapi "github.com/openshift/origin/pkg/deploy/api"
 )
 
+const (
+	// maxRetryCount is the number of times a deployment config will be retried before it is dropped
+	// out of the queue.
+	maxRetryCount = 15
+)
+
 // DeploymentTriggerController processes all triggers for a deployment config
 // and kicks new deployments whenever possible.
 type DeploymentTriggerController struct {
@@ -56,12 +62,13 @@ func (c *DeploymentTriggerController) handleErr(err error, key interface{}) {
 		return
 	}
 
-	if c.queue.NumRequeues(key) < MaxRetries {
+	if c.queue.NumRequeues(key) < maxRetryCount {
 		glog.V(2).Infof("Error instantiating deployment config %v: %v", key, err)
 		c.queue.AddRateLimited(key)
 		return
 	}
 
 	utilruntime.HandleError(err)
+	glog.V(2).Infof("Dropping deployment config %q out of the trigger queue: %v", key, err)
 	c.queue.Forget(key)
 }

pkg/deploy/controller/generictrigger/factory.go

@@ -24,9 +24,6 @@ const (
 	// stream stores have synced. If it hasn't synced, to avoid a hot loop, we'll wait this long
 	// between checks.
 	storeSyncedPollPeriod = 100 * time.Millisecond
-	// MaxRetries is the number of times a deployment config will be retried before it is dropped
-	// out of the queue.
-	MaxRetries = 5
 )
 
 // NewDeploymentTriggerController returns a new DeploymentTriggerController.