func setupTestSchedulerLongBindingWithRetry(queuedPodStore *clientcache.FIFO, scache schedulercache.Cache, nodeLister algorithm.FakeNodeLister, predicateMap map[string]algorithm.FitPredicate, stop chan struct{}, bindingTime time.Duration) (*Scheduler, chan *v1.Binding) {
algo := NewGenericScheduler(
scache,
predicateMap,
algorithm.EmptyMetadataProducer,
[]algorithm.PriorityConfig{},
algorithm.EmptyMetadataProducer,
[]algorithm.SchedulerExtender{})
bindingChan := make(chan *v1.Binding, 2)
cfg := &Config{
SchedulerCache: scache,
NodeLister: nodeLister,
Algorithm: algo,
Binder: fakeBinder{func(b *v1.Binding) error {
time.Sleep(bindingTime)
bindingChan <- b
return nil
}},
NextPod: func() *v1.Pod {
return clientcache.Pop(queuedPodStore).(*v1.Pod)
},
Error: func(p *v1.Pod, err error) {
queuedPodStore.AddIfNotPresent(p)
},
Recorder: &record.FakeRecorder{},
PodConditionUpdater: fakePodConditionUpdater{},
StopEverything: stop,
}
return New(cfg), bindingChan
}
func (factory *ConfigFactory) makeDefaultErrorFunc(backoff *podBackoff, podQueue *cache.FIFO) func(pod *v1.Pod, err error) {
return func(pod *v1.Pod, err error) {
if err == scheduler.ErrNoNodesAvailable {
glog.V(4).Infof("Unable to schedule %v %v: no nodes are registered to the cluster; waiting", pod.Namespace, pod.Name)
} else {
glog.Errorf("Error scheduling %v %v: %v; retrying", pod.Namespace, pod.Name, err)
}
backoff.gc()
// Retry asynchronously.
// Note that this is extremely rudimentary and we need a more real error handling path.
go func() {
defer runtime.HandleCrash()
podID := types.NamespacedName{
Namespace: pod.Namespace,
Name: pod.Name,
}
entry := backoff.getEntry(podID)
if !entry.TryWait(backoff.maxDuration) {
glog.Warningf("Request for pod %v already in flight, abandoning", podID)
return
}
// Get the pod again; it may have changed/been scheduled already.
getBackoff := initialGetBackoff
for {
pod, err := factory.Client.Core().Pods(podID.Namespace).Get(podID.Name, metav1.GetOptions{})
if err == nil {
if len(pod.Spec.NodeName) == 0 {
podQueue.AddIfNotPresent(pod)
}
break
}
if errors.IsNotFound(err) {
glog.Warningf("A pod %v no longer exists", podID)
return
}
glog.Errorf("Error getting pod %v for retry: %v; retrying...", podID, err)
if getBackoff = getBackoff * 2; getBackoff > maximalGetBackoff {
getBackoff = maximalGetBackoff
}
time.Sleep(getBackoff)
}
}()
}
}
func (factory *ConfigFactory) makeDefaultErrorFunc(backoff *podBackoff, podQueue *cache.FIFO) func(pod *api.Pod, err error) {
return func(pod *api.Pod, err error) {
if err == scheduler.ErrNoNodesAvailable {
glog.V(4).Infof("Unable to schedule %v %v: no nodes are registered to the cluster; waiting", pod.Namespace, pod.Name)
} else {
glog.Errorf("Error scheduling %v %v: %v; retrying", pod.Namespace, pod.Name, err)
}
backoff.gc()
// Retry asynchronously.
// Note that this is extremely rudimentary and we need a more real error handling path.
go func() {
defer runtime.HandleCrash()
podID := types.NamespacedName{
Namespace: pod.Namespace,
Name: pod.Name,
}
entry := backoff.getEntry(podID)
if !entry.TryWait(backoff.maxDuration) {
glog.Warningf("Request for pod %v already in flight, abandoning", podID)
return
}
// Get the pod again; it may have changed/been scheduled already.
pod = &api.Pod{}
err := factory.Client.Get().Namespace(podID.Namespace).Resource("pods").Name(podID.Name).Do().Into(pod)
if err != nil {
if !errors.IsNotFound(err) {
glog.Errorf("Error getting pod %v for retry: %v; abandoning", podID, err)
}
return
}
if pod.Spec.NodeName == "" {
podQueue.AddIfNotPresent(pod)
}
}()
}
}
