Gang

Prerequisites

• A cluster of v1.19 or later is available. For details, see Buying a CCE Cluster.
• The Volcano add-on has been installed. For details, see Volcano Scheduler.

Features

The Gang scheduling policy is one of the core scheduling algorithms of Volcano. It meets the scheduling requirements of "All or nothing" in the scheduling process and avoids the waste of cluster resources caused by arbitrary scheduling of pods. The Gang scheduler algorithm checks whether the number of scheduled pods in a job meets the minimum requirements for running the job. If yes, all pods in the job will be scheduled. If no, the pods will not be scheduled.

The Gang scheduling algorithm based on container groups is well suitable for scenarios where multi-process collaboration is required. AI scenarios typically involve complex processes. Data ingestion, data analysts, data splitting, trainers, serving, and logging which require a group of containers to work together are suitable for container-based Gang scheduling. Multi-thread parallel computing communication scenarios under MPI computing framework are also suitable for Gang scheduling because master and slave processes need to work together. Containers in a pod group are highly correlated, and there may be resource contention. The overall scheduling allocation can effectively resolve deadlocks. If cluster resources are insufficient, Gang scheduling can significantly improve the utilization of cluster resources.

Procedure

After Volcano is installed, you can enable or disable Gang scheduling on the Scheduling Configuration page. This function is enabled by default.

1. Log in to the CCE console.
2. Click the cluster name to access the cluster console. Choose Settings in the navigation pane and click the Scheduling Configuration tab.
3. In the AI task performance enhanced scheduling pane, select whether to enable Gang.

   This function helps you enhance the service throughput of the cluster and improve service running performance.

4. Click Confirm.
5. After the configuration, use Gang scheduling in workloads or Volcano jobs.
   • Create a workload using Gang scheduling.
     1. Create a pod group and specify minMember and minResources as follows:

        apiVersion: scheduling.volcano.sh/v1beta1
        kind: PodGroup
        metadata:
          name: pg-test1
        spec:
          minMember: 3
          minResources:
            cpu: 3
            memory: 3Gi

        • minMember: specifies the minimum requirement on the number of pods for running a workload. When the number of pods in the current pod group meets the requirement, these pods can be centrally scheduled.
        • minResources: specifies the minimum requirement on resources for running a workload. When the available resources in a cluster meet the requirement, the group of pods can be centrally scheduled.
     2. When creating a workload, use schedulerName to specify the Volcano scheduler and annotation to specify the pod group in which the Volcano scheduler runs.

        apiVersion: apps/v1
        kind: Deployment
        metadata:
          name: podgroup-test
          labels:
            app: podgroup-test
        spec:
          replicas: 6
          selector:
            matchLabels:
              app: podgroup-test
          template:
            metadata:
              annotations:
                scheduling.k8s.io/group-name: pg-test1
              labels:
                app: podgroup-test
            spec:
              schedulerName: volcano
              containers:
              - name: test
                image: busybox
                imagePullPolicy: IfNotPresent
                command: ['sh', '-c', 'echo "Hello, Kubernetes!" && sleep 3600']
                resources:
                  requests:
                    cpu: 500m
                  limits:
                    cpu: 500m

        • schedulerName: Set this parameter to volcano, indicating that Volcano will be used to schedule pods for the workload.
        • scheduling.k8s.io/group-name: specifies the pod group created in the previous step, for example, pg-test1.
   • Create a Volcano job using Gang scheduling.

     When creating a Volcano job, you only need to configure minAvailable and set schedulerName to volcano. The Volcano scheduler will automatically create a pod group and manage it. The following shows an example:

     apiVersion: batch.volcano.sh/v1alpha1
     kind: Job
     metadata:
       name: vcjob
     spec:
       schedulerName: volcano
       minAvailable: 2
       tasks:
         - replicas: 4
           name: "test"
           template:
             spec:
               containers:
                 - image: alpine
                   command: ["/bin/sh", "-c", "sleep 1000"]
                   imagePullPolicy: IfNotPresent
                   name: running
                   resources:
                     requests:
                       cpu: "1"
               restartPolicy: OnFailure