github.com/kubewharf/katalyst-core@v0.5.3/pkg/agent/orm/checkpoint.go (about)

     1  /*
     2  Copyright 2022 The Katalyst Authors.
     3  
     4  Licensed under the Apache License, Version 2.0 (the "License");
     5  you may not use this file except in compliance with the License.
     6  You may obtain a copy of the License at
     7  
     8      http://www.apache.org/licenses/LICENSE-2.0
     9  
    10  Unless required by applicable law or agreed to in writing, software
    11  distributed under the License is distributed on an "AS IS" BASIS,
    12  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    13  See the License for the specific language governing permissions and
    14  limitations under the License.
    15  */
    16  
    17  package orm
    18  
    19  import (
    20  	"fmt"
    21  	"path/filepath"
    22  
    23  	"k8s.io/klog/v2"
    24  	"k8s.io/kubernetes/pkg/kubelet/checkpointmanager/errors"
    25  
    26  	"github.com/kubewharf/katalyst-core/pkg/agent/orm/checkpoint"
    27  	"github.com/kubewharf/katalyst-core/pkg/agent/orm/endpoint"
    28  	"github.com/kubewharf/katalyst-core/pkg/consts"
    29  )
    30  
    31  func (m *ManagerImpl) checkpointFile() string {
    32  	return filepath.Join(m.socketdir, consts.KubeletQoSResourceManagerCheckpoint)
    33  }
    34  
    35  func (m *ManagerImpl) writeCheckpoint() error {
    36  	data := checkpoint.New(m.podResources.toCheckpointData())
    37  	err := m.checkpointManager.CreateCheckpoint(consts.KubeletQoSResourceManagerCheckpoint, data)
    38  	if err != nil {
    39  		err = fmt.Errorf("[ORM] failed to write checkpoint file %q: %v", consts.KubeletQoSResourceManagerCheckpoint, err)
    40  		klog.Warning(err)
    41  		return err
    42  	}
    43  	return nil
    44  }
    45  
    46  func (m *ManagerImpl) readCheckpoint() error {
    47  	resEntries := make([]checkpoint.PodResourcesEntry, 0)
    48  	cp := checkpoint.New(resEntries)
    49  	err := m.checkpointManager.GetCheckpoint(consts.KubeletQoSResourceManagerCheckpoint, cp)
    50  	if err != nil {
    51  		if err == errors.ErrCheckpointNotFound {
    52  			klog.Warningf("[ORM] Failed to retrieve checkpoint for %q: %v", consts.KubeletQoSResourceManagerCheckpoint, err)
    53  			return nil
    54  		}
    55  		return err
    56  	}
    57  
    58  	podResources := cp.GetData()
    59  	klog.V(5).Infof("[ORM] read checkpoint %v", podResources)
    60  	m.podResources.fromCheckpointData(podResources)
    61  
    62  	m.mutex.Lock()
    63  
    64  	allocatedResourceNames := m.podResources.allAllocatedResourceNames()
    65  
    66  	for _, allocatedResourceName := range allocatedResourceNames.UnsortedList() {
    67  		m.endpoints[allocatedResourceName] = endpoint.EndpointInfo{E: endpoint.NewStoppedEndpointImpl(allocatedResourceName), Opts: nil}
    68  	}
    69  
    70  	m.mutex.Unlock()
    71  
    72  	return nil
    73  }