Uploaded image for project: 'Apache YuniKorn'
  1. Apache YuniKorn
  2. YUNIKORN-1165

Yunikorn plugin inconsistent node assignment + scheduling

    XMLWordPrintableJSON

Details

    Description

      Although the plugin selects a certain node, the default scheduler may not schedule the pod on the plugin-selected node. In the below example, Yunikorn selected node host-05.shared-ek8s-dev-01.kcloud.cloudera.com. However, the pod was scheduled on host-01.shared-ek8s-dev-01.kcloud.cloudera.com.

      Internally, Yunikorn still incorrectly believes that the node was scheduled on host-05.shared-ek8s-dev-01.kcloud.cloudera.com.

      Events:
        Type     Reason            Age   From      Message
        ----     ------            ----  ----      -------
        Warning  FailedScheduling  19s   yunikorn  0/10 nodes are available: 10 Pod is not ready for scheduling.
        Warning  FailedScheduling  19s   yunikorn  0/10 nodes are available: 10 Pod is not ready for scheduling.
        Normal   Scheduling        16s   yunikorn  dev/test-pod-1 is queued and waiting for allocation
        Normal   QuotaApproved     16s   yunikorn  Pod dev/test-pod-1 is ready for scheduling on node host-05.shared-ek8s-dev-01.kcloud.cloudera.com
        Normal   Scheduled         16s   yunikorn  Successfully assigned dev/test-pod-1 to host-01.shared-ek8s-dev-01.kcloud.cloudera.com
        Normal   Pulled            15s   kubelet   Container image "docker-private.infra.cloudera.com/cloudera_base/alpine:latest" already present on machine
        Normal   Created           15s   kubelet   Created container sleep-600s-1
        Normal   Started           15s   kubelet   Started container sleep-600s-1 
      {         "applicationID": "app-v1",         "usedResource": "[memory:50 vcore:50]",         "partition": "default",         "queueName": "root.dev",         "submissionTime": 1648676490424350638,         "allocations":         [             {                 "allocationKey": "d50084ea-2ae1-4057-8a9b-48cda1fad4c7",                 "allocationTags":                 {                     "kubernetes.io/label/app": "sleep",                     "kubernetes.io/label/applicationId": "app-v1",                     "kubernetes.io/label/component": "yunikorn-scheduler",                     "kubernetes.io/label/queue": "root.default",                     "kubernetes.io/meta/namespace": "dev",                     "kubernetes.io/meta/podName": "test-pod-1"                 },                 "uuid": "f7e4b5d6-1e71-4d5b-bd1d-aa28067a3d4b",                 "resource": "[memory:50 vcore:50]",                 "priority": "0",                 "queueName": "root.dev",                 "nodeId": "host-05.shared-ek8s-dev-01.kcloud.cloudera.com",                 "applicationId": "app-v1",                 "partition": "default"             }         ],         "applicationState": "Starting"     },
      

      Attachments

        Issue Links

          Activity

            People

              ccondit Craig Condit
              rozhang Ronald Zhang
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: