This repository was archived by the owner on Apr 24, 2023. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 42
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
mark apps that exceed the cluster capacity with a custom pod condition (
#35) * mark apps that exceed the cluster capacity with a custom pod condition * address comments * move params * less logging duplication
- Loading branch information
1 parent
af85a36
commit 075e989
Showing
5 changed files
with
176 additions
and
7 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,154 @@ | ||
| // Copyright (c) 2019 Palantir Technologies. All rights reserved. | ||
| // | ||
| // Licensed under the Apache License, Version 2.0 (the "License"); | ||
| // you may not use this file except in compliance with the License. | ||
| // You may obtain a copy of the License at | ||
| // | ||
| // http://www.apache.org/licenses/LICENSE-2.0 | ||
| // | ||
| // Unless required by applicable law or agreed to in writing, software | ||
| // distributed under the License is distributed on an "AS IS" BASIS, | ||
| // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
| // See the License for the specific language governing permissions and | ||
| // limitations under the License. | ||
|
|
||
| package extender | ||
|
|
||
| import ( | ||
| "context" | ||
| "time" | ||
|
|
||
| "github.com/palantir/k8s-spark-scheduler-lib/pkg/resources" | ||
| "github.com/palantir/witchcraft-go-logging/wlog/svclog/svc1log" | ||
| "github.com/palantir/witchcraft-go-logging/wlog/wapp" | ||
| v1 "k8s.io/api/core/v1" | ||
| "k8s.io/apimachinery/pkg/labels" | ||
| corev1 "k8s.io/client-go/kubernetes/typed/core/v1" | ||
| corelisters "k8s.io/client-go/listers/core/v1" | ||
| podutil "k8s.io/kubernetes/pkg/api/v1/pod" | ||
| ) | ||
|
|
||
| const ( | ||
| podExceedsClusterCapacity v1.PodConditionType = "PodExceedsClusterCapacity" | ||
| unschedulablePollingInterval time.Duration = time.Minute | ||
| unschedulableInClusterThreshold time.Duration = 10 * time.Minute | ||
| ) | ||
|
|
||
| // UnschedulablePodMarker checks for spark scheduler managed pending driver pods | ||
| // and checks if they can fit if the cluster was empty, else marks them with a | ||
| // custom pod condition. | ||
| type UnschedulablePodMarker struct { | ||
| nodeLister corelisters.NodeLister | ||
| podLister corelisters.PodLister | ||
| coreClient corev1.CoreV1Interface | ||
| overheadComputer *OverheadComputer | ||
| binpacker *Binpacker | ||
| } | ||
|
|
||
| // NewUnschedulablePodMarker creates a new UnschedulablePodMarker | ||
| func NewUnschedulablePodMarker( | ||
| nodeLister corelisters.NodeLister, | ||
| podLister corelisters.PodLister, | ||
| coreClient corev1.CoreV1Interface, | ||
| overheadComputer *OverheadComputer, | ||
| binpacker *Binpacker) *UnschedulablePodMarker { | ||
| return &UnschedulablePodMarker{ | ||
| nodeLister: nodeLister, | ||
| podLister: podLister, | ||
| coreClient: coreClient, | ||
| overheadComputer: overheadComputer, | ||
| binpacker: binpacker, | ||
| } | ||
| } | ||
|
|
||
| // Start starts periodic scanning for unschedulable applications | ||
| func (u *UnschedulablePodMarker) Start(ctx context.Context) { | ||
| _ = wapp.RunWithFatalLogging(ctx, u.doStart) | ||
| } | ||
|
|
||
| func (u *UnschedulablePodMarker) doStart(ctx context.Context) error { | ||
| t := time.NewTicker(unschedulablePollingInterval) | ||
| for { | ||
| select { | ||
| case <-ctx.Done(): | ||
| return nil | ||
| case <-t.C: | ||
| u.scanForUnschedulablePods(ctx) | ||
| } | ||
| } | ||
| } | ||
|
|
||
| func (u *UnschedulablePodMarker) scanForUnschedulablePods(ctx context.Context) { | ||
| pods, err := u.podLister.List(labels.Everything()) | ||
| if err != nil { | ||
| svc1log.FromContext(ctx).Error("failed to list pods", svc1log.Stacktrace(err)) | ||
| return | ||
| } | ||
| now := time.Now() | ||
| for _, pod := range pods { | ||
| if pod.Spec.SchedulerName == SparkSchedulerName && | ||
| len(pod.Spec.NodeName) == 0 && | ||
| pod.DeletionTimestamp == nil && | ||
| pod.Labels[SparkRoleLabel] == Driver && | ||
| pod.CreationTimestamp.Time.Add(unschedulableInClusterThreshold).Before(now) { | ||
|
|
||
| ctx = svc1log.WithLoggerParams( | ||
| ctx, | ||
| svc1log.SafeParam("podName", pod.Name), | ||
| svc1log.SafeParam("podNamespace", pod.Namespace)) | ||
|
|
||
| exceedsCapacity, err := u.doesPodExceedClusterCapacity(ctx, pod) | ||
| if err != nil { | ||
| svc1log.FromContext(ctx).Error("failed to check if pod was unschedulable", | ||
| svc1log.Stacktrace(err)) | ||
| return | ||
| } | ||
|
|
||
| err = u.markPodClusterCapacityStatus(ctx, pod, exceedsCapacity) | ||
| if err != nil { | ||
| svc1log.FromContext(ctx).Error("failed to mark pod cluster capacity status", | ||
| svc1log.Stacktrace(err)) | ||
| } | ||
| } | ||
| } | ||
| } | ||
|
|
||
| func (u *UnschedulablePodMarker) doesPodExceedClusterCapacity(ctx context.Context, driver *v1.Pod) (bool, error) { | ||
| nodes, err := u.nodeLister.List(labels.Set(driver.Spec.NodeSelector).AsSelector()) | ||
| if err != nil { | ||
| return false, err | ||
| } | ||
| nodeNames := make([]string, 0, len(nodes)) | ||
| for _, node := range nodes { | ||
| nodeNames = append(nodeNames, node.Name) | ||
| } | ||
| availableResources := resources.AvailableForNodes(nodes, u.overheadComputer.GetOverhead(ctx, nodes)) | ||
| applicationResources, err := sparkResources(ctx, driver) | ||
| if err != nil { | ||
| return false, err | ||
| } | ||
| _, _, hasCapacity := u.binpacker.BinpackFunc( | ||
| ctx, | ||
| applicationResources.driverResources, | ||
| applicationResources.executorResources, | ||
| applicationResources.executorCount, | ||
| nodeNames, | ||
| nodeNames, | ||
| availableResources) | ||
|
|
||
| return !hasCapacity, nil | ||
| } | ||
|
|
||
| func (u *UnschedulablePodMarker) markPodClusterCapacityStatus(ctx context.Context, driver *v1.Pod, exceedsCapacity bool) error { | ||
| exceedsCapacityStatus := v1.ConditionFalse | ||
| if exceedsCapacity { | ||
| exceedsCapacityStatus = v1.ConditionTrue | ||
| } | ||
|
|
||
| if !podutil.UpdatePodCondition(&driver.Status, &v1.PodCondition{Type: podExceedsClusterCapacity, Status: exceedsCapacityStatus}) { | ||
| return nil | ||
| } | ||
|
|
||
| _, err := u.coreClient.Pods(driver.Namespace).UpdateStatus(driver) | ||
| return err | ||
| } |