fix(helm): No more retry if jobs failed beyond BackoffLimit

Signed-off-by: dayeguilaiye <979014041@qq.com>

pkg/kube/ready.go

@@ -18,7 +18,7 @@ package kube // import "helm.sh/helm/v3/pkg/kube"
 
 import (
 	"context"
-	"fmt"
+	"github.com/pkg/errors"
 
 	appsv1 "k8s.io/api/apps/v1"
 	appsv1beta1 "k8s.io/api/apps/v1beta1"
@@ -241,7 +241,7 @@ func (c *ReadyChecker) jobReady(job *batchv1.Job) (bool, error) {
 	if job.Status.Failed > *job.Spec.BackoffLimit {
 		c.log("Job is failed: %s/%s", job.GetNamespace(), job.GetName())
 		// If a job is failed, it can't recover, so throw an error
-		return false, fmt.Errorf("job is failed: %s/%s", job.GetNamespace(), job.GetName())
+		return false, errors.Wrapf(ErrNoRetryError, "job is failed: %s/%s", job.GetNamespace(), job.GetName())
 	}
 	if job.Spec.Completions != nil && job.Status.Succeeded < *job.Spec.Completions {
 		c.log("Job is not completed: %s/%s", job.GetNamespace(), job.GetName())

pkg/kube/wait.go

@@ -38,6 +38,8 @@ import (
 	"k8s.io/apimachinery/pkg/util/wait"
 )
 
+var ErrNoRetryError = errors.New("this error will stop retry")
+
 type waiter struct {
 	c       ReadyChecker
 	timeout time.Duration
@@ -91,6 +93,10 @@ func (w *waiter) isRetryableError(err error, resource *resource.Info) bool {
 		w.log("Status code received: %d. Retryable error? %t", statusCode, retryable)
 		return retryable
 	}
+	if errors.Is(err, ErrNoRetryError) {
+		w.log("The error is a NoRetryError, Retryable err? %t", false)
+		return false
+	}
 	w.log("Retryable error? %t", true)
 	return true
 }