Add support for basic cgroup management

pkg/kubelet/cm/cgroup_manager_linux.go

@@ -0,0 +1,113 @@
+/*
+Copyright 2016 The Kubernetes Authors 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 cm
+
+import (
+	"fmt"
+
+	libcontainerconfigs "github.com/opencontainers/runc/libcontainer/configs"
+)
+
+// cgroupManagerImpl implements the CgroupManager interface.
+// Its a stateless object which can be used to
+// update,create or delete any number of cgroups
+// It uses the Libcontainer raw fs cgroup manager for cgroup management.
+type cgroupManagerImpl struct {
+	// subsystems holds information about all the
+	// mounted cgroup subsytems on the node
+	subsystems *cgroupSubsystems
+}
+
+// Make sure that cgroupManagerImpl implements the CgroupManager interface
+var _ CgroupManager = &cgroupManagerImpl{}
+
+// NewCgroupManager is a factory method that returns a CgroupManager
+func NewCgroupManager(cs *cgroupSubsystems) CgroupManager {
+	return &cgroupManagerImpl{
+		subsystems: cs,
+	}
+}
+
+// Destroy destroys the specified cgroup
+func (m *cgroupManagerImpl) Destroy(cgroupConfig *CgroupConfig) error {
+	//cgroup name
+	name := cgroupConfig.Name
+
+	// get the fscgroup Manager with the specified cgroup configuration
+	fsCgroupManager, err := getLibcontainerCgroupManager(cgroupConfig, m.subsystems)
+
+	if err != nil {
+		return fmt.Errorf("Unable to destroy cgroup paths for cgroup %v : %v", name, err)
+	}
+	// Delete cgroups using libcontainers Managers Destroy() method
+	if err := fsCgroupManager.Destroy(); err != nil {
+		return fmt.Errorf("Unable to destroy cgroup paths for cgroup %v : %v", name, err)
+	}
+	return nil
+}
+
+// Update updates the cgroup with the specified Cgroup Configuration
+func (m *cgroupManagerImpl) Update(cgroupConfig *CgroupConfig) error {
+	//cgroup name
+	name := cgroupConfig.Name
+
+	// get the fscgroup Manager with the specified cgroup configuration
+	fsCgroupManager, err := getLibcontainerCgroupManager(cgroupConfig, m.subsystems)
+	if err != nil {
+		return fmt.Errorf("Failed to update cgroup for %v : %v", name, err)
+	}
+	// get config object for passing to Set()
+	config := &libcontainerconfigs.Config{
+		Cgroups: fsCgroupManager.Cgroups,
+	}
+
+	// Update cgroup configuration using libcontainers Managers Set() method
+	if err := fsCgroupManager.Set(config); err != nil {
+		return fmt.Errorf("Failed to update cgroup for %v: %v", name, err)
+	}
+	return nil
+}
+
+// Create creates the specified cgroup
+func (m *cgroupManagerImpl) Create(cgroupConfig *CgroupConfig) error {
+	//cgroup name
+	name := cgroupConfig.Name
+
+	// get the fscgroup Manager with the specified cgroup configuration
+	fsCgroupManager, err := getLibcontainerCgroupManager(cgroupConfig, m.subsystems)
+	if err != nil {
+		return fmt.Errorf("Failed to create cgroup for %v : %v", name, err)
+	}
+	// get config object for passing to libcontainer's Set() method
+	config := &libcontainerconfigs.Config{
+		Cgroups: fsCgroupManager.Cgroups,
+	}
+	//Apply(0) is a hack to create the cgroup directories for each resource
+	// subsystem. The function [cgroups.Manager.apply()] applies cgroup
+	// configuration to the process with the specified pid.
+	// It creates cgroup files for each subsytems and writes the pid
+	// in the tasks file. We use the function to create all the required
+	// cgroup files but not attach any "real" pid to the cgroup.
+	if err := fsCgroupManager.Apply(0); err != nil {
+		return fmt.Errorf("Failed to create cgroup for %v: %v", name, err)
+	}
+	// Update cgroup configuration using libcontainers Managers Set() method
+	if err := fsCgroupManager.Set(config); err != nil {
+		return fmt.Errorf("Failed to create cgroup for %v: %v", name, err)
+	}
+	return nil
+}

pkg/kubelet/cm/cgroup_manager_unsupported.go

@@ -0,0 +1,42 @@
+// +build !linux
+
+/*
+Copyright 2016 The Kubernetes Authors 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 cm
+
+import "fmt"
+
+type unsupportedCgroupManager struct{}
+
+// Make sure that unsupportedCgroupManager implements the CgroupManager interface
+var _ CgroupManager = &unsupportedCgroupManager{}
+
+func NewCgroupManager(_ *cgroupSubsystems) CgroupManager {
+	return &unsupportedCgroupManager{}
+}
+
+func (m *unsupportedCgroupManager) Destroy(_ *CgroupConfig) error {
+	return nil
+}
+
+func (m *unsupportedCgroupManager) Update(_ *CgroupConfig) error {
+	return nil
+}
+
+func (m *unsupportedCgroupManager) Create(_ *CgroupConfig) error {
+	return fmt.Errorf("Cgroup Manager is not supported in this build")
+}

pkg/kubelet/cm/helpers_linux.go

@@ -0,0 +1,98 @@
+/*
+Copyright 2016 The Kubernetes Authors 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 cm
+
+import (
+	"fmt"
+	"path"
+
+	libcontainercgroups "github.com/opencontainers/runc/libcontainer/cgroups"
+	cgroupfs "github.com/opencontainers/runc/libcontainer/cgroups/fs"
+	libcontainerconfigs "github.com/opencontainers/runc/libcontainer/configs"
+)
+
+// cgroupSubsystems holds information about the mounted cgroup subsytems
+type cgroupSubsystems struct {
+	// Cgroup subsystem mounts.
+	// e.g.: "/sys/fs/cgroup/cpu" -> ["cpu", "cpuacct"]
+	mounts []libcontainercgroups.Mount
+
+	// Cgroup subsystem to their mount location.
+	// e.g.: "cpu" -> "/sys/fs/cgroup/cpu"
+	mountPoints map[string]string
+}
+
+// GetCgroupSubsystems returns information about the mounted cgroup subsystems
+func getCgroupSubsystems() (*cgroupSubsystems, error) {
+	// Get all cgroup mounts.
+	allCgroups, err := libcontainercgroups.GetCgroupMounts()
+	if err != nil {
+		return &cgroupSubsystems{}, err
+	}
+	if len(allCgroups) == 0 {
+		return &cgroupSubsystems{}, fmt.Errorf("failed to find cgroup mounts")
+	}
+
+	//TODO(@dubstack) should we trim to only the supported ones
+	mountPoints := make(map[string]string, len(allCgroups))
+	for _, mount := range allCgroups {
+		for _, subsystem := range mount.Subsystems {
+			mountPoints[subsystem] = mount.Mountpoint
+		}
+	}
+	return &cgroupSubsystems{
+		mounts:      allCgroups,
+		mountPoints: mountPoints,
+	}, nil
+}
+
+// getLibcontainerCgroupManager returns libcontainer's cgroups manager
+// object with the specified cgroup configuration
+func getLibcontainerCgroupManager(cgroupConfig *CgroupConfig, subsystems *cgroupSubsystems) (*cgroupfs.Manager, error) {
+	// get cgroup name
+	name := cgroupConfig.Name
+
+	// Get map of all cgroup paths on the system for the particular cgroup
+	cgroupPaths := make(map[string]string, len(subsystems.mountPoints))
+	for key, val := range subsystems.mountPoints {
+		cgroupPaths[key] = path.Join(val, name)
+	}
+
+	// Extract the cgroup resource parameters
+	resourceConfig := cgroupConfig.ResourceParameters
+	resources := &libcontainerconfigs.Resources{}
+	resources.AllowAllDevices = true
+	if resourceConfig.Memory != nil {
+		resources.Memory = *resourceConfig.Memory
+	}
+	if resourceConfig.CpuShares != nil {
+		resources.CpuShares = *resourceConfig.CpuShares
+	}
+	if resourceConfig.CpuQuota != nil {
+		resources.CpuQuota = *resourceConfig.CpuQuota
+	}
+	// Initialize libcontainer's cgroup config
+	libcontainerCgroupConfig := &libcontainerconfigs.Cgroup{
+		Name:      path.Base(name),
+		Parent:    path.Dir(name),
+		Resources: resources,
+	}
+	return &cgroupfs.Manager{
+		Cgroups: libcontainerCgroupConfig,
+		Paths:   cgroupPaths,
+	}, nil
+}

pkg/kubelet/cm/types.go

@@ -0,0 +1,56 @@
+/*
+Copyright 2016 The Kubernetes Authors 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 cm
+
+// ResourceConfig holds information about all the supported cgroup resource parameters.
+type ResourceConfig struct {
+	// Memory limit (in bytes).
+	Memory *int64
+	// CPU shares (relative weight vs. other containers).
+	CpuShares *int64
+	// CPU hardcap limit (in usecs). Allowed cpu time in a given period.
+	CpuQuota *int64
+}
+
+// CgroupConfig holds the cgroup configuration information.
+// This is common object which is used to specify
+// cgroup information to both systemd and raw cgroup fs
+// implementation of the Cgroup Manager interface.
+type CgroupConfig struct {
+	// We would expect systemd implementation to make appropriate
+	// name conversion. For example, if we pass /foo/bar
+	// then systemd should convert the name to something like
+	// foo.slice/foo-bar.slice
+
+	// Fully qualified name
+	Name string
+	// ResourceParameters contains various cgroups settings to apply.
+	ResourceParameters *ResourceConfig
+}
+
+// CgroupManager allows for cgroup management.
+// Supports Cgroup Creation ,Deletion and Updates.
+type CgroupManager interface {
+	// Create creates and applies the cgroup configurations on the cgroup.
+	// It just creates the leaf cgroups.
+	// It expects the parent cgroup to already exist.
+	Create(*CgroupConfig) error
+	// Destroys the cgroup.
+	Destroy(*CgroupConfig) error
+	// Update cgroup configuration.
+	Update(*CgroupConfig) error
+}