// Copyright 2014 Google Inc. 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 docker import ( "flag" "fmt" "path" "regexp" "strconv" "strings" "sync" "github.com/docker/libcontainer/cgroups" "github.com/fsouza/go-dockerclient" "github.com/golang/glog" "github.com/google/cadvisor/container" "github.com/google/cadvisor/container/libcontainer" "github.com/google/cadvisor/fs" info "github.com/google/cadvisor/info/v1" "github.com/google/cadvisor/utils" ) var ArgDockerEndpoint = flag.String("docker", "unix:///var/run/docker.sock", "docker endpoint") // The namespace under which Docker aliases are unique. var DockerNamespace = "docker" // Basepath to all container specific information that libcontainer stores. var dockerRootDir = flag.String("docker_root", "/var/lib/docker", "Absolute path to the Docker state root directory (default: /var/lib/docker)") var dockerRunDir = flag.String("docker_run", "/var/run/docker", "Absolute path to the Docker run directory (default: /var/run/docker)") // Regexp that identifies docker cgroups, containers started with // --cgroup-parent have another prefix than 'docker' var dockerCgroupRegexp = regexp.MustCompile(`.+-([a-z0-9]{64})\.scope$`) // TODO(vmarmol): Export run dir too for newer Dockers. // Directory holding Docker container state information. func DockerStateDir() string { return libcontainer.DockerStateDir(*dockerRootDir) } // Whether the system is using Systemd. var useSystemd bool var check = sync.Once{} func UseSystemd() bool { check.Do(func() { useSystemd = false // Check for system.slice in systemd and cpu cgroup. for _, cgroupType := range []string{"name=systemd", "cpu"} { mnt, err := cgroups.FindCgroupMountpoint(cgroupType) if err == nil { // systemd presence does not mean systemd controls cgroups. // If system.slice cgroup exists, then systemd is taking control. // This breaks if user creates system.slice manually :) if utils.FileExists(path.Join(mnt, "system.slice")) { useSystemd = true break } } } }) return useSystemd } func RootDir() string { return *dockerRootDir } type storageDriver string const ( devicemapperStorageDriver storageDriver = "devicemapper" aufsStorageDriver storageDriver = "aufs" ) type dockerFactory struct { machineInfoFactory info.MachineInfoFactory storageDriver storageDriver client *docker.Client // Information about the mounted cgroup subsystems. cgroupSubsystems libcontainer.CgroupSubsystems // Information about mounted filesystems. fsInfo fs.FsInfo } func (self *dockerFactory) String() string { return DockerNamespace } func (self *dockerFactory) NewContainerHandler(name string, inHostNamespace bool) (handler container.ContainerHandler, err error) { client, err := Client() if err != nil { return } handler, err = newDockerContainerHandler( client, name, self.machineInfoFactory, self.fsInfo, self.storageDriver, &self.cgroupSubsystems, inHostNamespace, ) return } // Returns the Docker ID from the full container name. func ContainerNameToDockerId(name string) string { id := path.Base(name) // Turn systemd cgroup name into Docker ID. if UseSystemd() { if matches := dockerCgroupRegexp.FindStringSubmatch(id); matches != nil { id = matches[1] } } return id } func isContainerName(name string) bool { if UseSystemd() { return dockerCgroupRegexp.MatchString(path.Base(name)) } return true } // Docker handles all containers under /docker func (self *dockerFactory) CanHandleAndAccept(name string) (bool, bool, error) { // docker factory accepts all containers it can handle. canAccept := true if !isContainerName(name) { return false, canAccept, fmt.Errorf("invalid container name") } // Check if the container is known to docker and it is active. id := ContainerNameToDockerId(name) // We assume that if Inspect fails then the container is not known to docker. ctnr, err := self.client.InspectContainer(id) if err != nil || !ctnr.State.Running { return false, canAccept, fmt.Errorf("error inspecting container: %v", err) } return true, canAccept, nil } func (self *dockerFactory) DebugInfo() map[string][]string { return map[string][]string{} } func parseDockerVersion(full_version_string string) ([]int, error) { version_regexp_string := "(\\d+)\\.(\\d+)\\.(\\d+)" version_re := regexp.MustCompile(version_regexp_string) matches := version_re.FindAllStringSubmatch(full_version_string, -1) if len(matches) != 1 { return nil, fmt.Errorf("version string \"%v\" doesn't match expected regular expression: \"%v\"", full_version_string, version_regexp_string) } version_string_array := matches[0][1:] version_array := make([]int, 3) for index, version_string := range version_string_array { version, err := strconv.Atoi(version_string) if err != nil { return nil, fmt.Errorf("error while parsing \"%v\" in \"%v\"", version_string, full_version_string) } version_array[index] = version } return version_array, nil } // Register root container before running this function! func Register(factory info.MachineInfoFactory, fsInfo fs.FsInfo) error { if UseSystemd() { glog.Infof("System is using systemd") } client, err := Client() if err != nil { return fmt.Errorf("unable to communicate with docker daemon: %v", err) } if version, err := client.Version(); err != nil { return fmt.Errorf("unable to communicate with docker daemon: %v", err) } else { expected_version := []int{1, 0, 0} version_string := version.Get("Version") version, err := parseDockerVersion(version_string) if err != nil { return fmt.Errorf("couldn't parse docker version: %v", err) } for index, number := range version { if number > expected_version[index] { break } else if number < expected_version[index] { return fmt.Errorf("cAdvisor requires docker version %v or above but we have found version %v reported as \"%v\"", expected_version, version, version_string) } } } // Check that the libcontainer execdriver is used. information, err := DockerInfo() if err != nil { return fmt.Errorf("failed to detect Docker info: %v", err) } execDriver, ok := information["ExecutionDriver"] if !ok || !strings.HasPrefix(execDriver, "native") { return fmt.Errorf("docker found, but not using native exec driver") } sd, _ := information["Driver"] cgroupSubsystems, err := libcontainer.GetCgroupSubsystems() if err != nil { return fmt.Errorf("failed to get cgroup subsystems: %v", err) } glog.Infof("Registering Docker factory") f := &dockerFactory{ machineInfoFactory: factory, client: client, storageDriver: storageDriver(sd), cgroupSubsystems: cgroupSubsystems, fsInfo: fsInfo, } container.RegisterContainerHandlerFactory(f) return nil }