Kubelet源码分析(二): DockerClient

源码版本

kubernetes version： v1.3.0

DockerClient初始化

DockerClient是KubeletConfig的成员之一。
KubeletConfig结构介绍：

type KubeletConfig struct {
    Address                        net.IP
    AllowPrivileged                bool
...
    DockerClient                   dockertools.DockerInterface
    RuntimeCgroups                 string
    DockerExecHandler              dockertools.ExecHandler
...
}

而kubeletConfig的初始化是在UnsecuredKubeletConfig()接口中进行的，需要依赖最开始组建的kubeletServer配置结构，该kubeletServer结构中有DockerEndpoint字符串成员：

type KubeletServer struct {
    componentconfig.KubeletConfiguration

    AuthPath      util.StringFlag // Deprecated -- use KubeConfig instead
    KubeConfig    util.StringFlag
    APIServerList []string

    RunOnce bool

    // Insert a probability of random errors during calls to the master.
    ChaosChance float64
    // Crash immediately, rather than eating panics.
    ReallyCrashForTesting bool
    SystemReserved        config.ConfigurationMap
    KubeReserved          config.ConfigurationMap
}

type KubeletConfiguration struct {
    // config is the path to the config file or directory of files
    Config string `json:"config"`
...
    DockerEndpoint string `json:"dockerEndpoint"`
...    

实际上如果没有指定该参数的话，会默认使用端点"unix:///var/run/docker.sock"做为DockerEndpoint。可以查看NewEnvClient()接口。

回到kubeletConfig的初始化接口UnsecuredKubeletConfig()：

func UnsecuredKubeletConfig(s *options.KubeletServer) (*KubeletConfig, error) {
    hostNetworkSources, err := kubetypes.GetValidatedSources(strings.Split(s.HostNetworkSources, ","))
    if err != nil {
        return nil, err
    }
...
    return &KubeletConfig{
        Address:                      net.ParseIP(s.Address),
        AllowPrivileged:              s.AllowPrivileged,
        ...
        DockerClient:                 dockertools.ConnectToDockerOrDie(s.DockerEndpoint, s.RuntimeRequestTimeout.Duration), // TODO(random-liu): Set RuntimeRequestTimeout for rkt.
...
    }

接着继续查看dockertools.ConnectToDockerOrDie(s.DockerEndpoint, s.RuntimeRequestTimeout.Duration)。

func ConnectToDockerOrDie(dockerEndpoint string, requestTimeout time.Duration) DockerInterface {
    if dockerEndpoint == "fake://" {
        return NewFakeDockerClient()
    }
    client, err := getDockerClient(dockerEndpoint)
    if err != nil {
        glog.Fatalf("Couldn't connect to docker: %v", err)
    }
    glog.Infof("Start docker client with request timeout=%v", requestTimeout)
    return newKubeDockerClient(client, requestTimeout)
}

先前我们了解了如果在kubelet启动时没有传入"docker-endpoint"参数的话，s.DockerEndpoint即为空。
s.RuntimeRequestTimeout.Duration值可以查看NewKubeletServer()函数的初始化，是2min。
getDockerClient()接口比较简单：
getDockerClient --> dockerapi.NewEnvClient() --> NewClient().
NewClient()接口如下：

func NewClient(host string, version string, client *http.Client, httpHeaders map[string]string) (*Client, error) {
    proto, addr, basePath, err := ParseHost(host)
    if err != nil {
        return nil, err
    }

    transport, err := transport.NewTransportWithHTTP(proto, addr, client)
    if err != nil {
        return nil, err
    }

    return &Client{
        proto:             proto,
        addr:              addr,
        basePath:          basePath,
        transport:         transport,
        version:           version,
        customHTTPHeaders: httpHeaders,
    }, nil
}

之前讲了如果没有传入"docker-endpoint"参数的话，默认值就是"unix:///var/run/docker.sock".即host参数为该值。
ParseHost()先根据host进行解析，然后创建transport-->Client。
Client结构如下：

type Client struct {
    // proto holds the client protocol i.e. unix.
    proto string
    // addr holds the client address.
    addr string
    // basePath holds the path to prepend to the requests.
    basePath string
    // transport is the interface to send request with, it implements transport.Client.
    transport transport.Client
    // version of the server to talk to.
    version string
    // custom http headers configured by users.
    customHTTPHeaders map[string]string
}

创建Client成功之后，最终开始提到的ConnectToDockerOrDie()接口会调用newKubeDockerClient()生成pkg/kubelet/dockertools/kube_docker_client.go里的kubeDockerClient结构：

type kubeDockerClient struct {
    // timeout is the timeout of short running docker operations.
    timeout time.Duration
    client  *dockerapi.Client
}

初始化到这里就结束了，那我们回到最初，介绍下DockerClient定义：
dockertools.DockerInterface如下：

type DockerInterface interface {
    ListContainers(options dockertypes.ContainerListOptions) ([]dockertypes.Container, error)
    InspectContainer(id string) (*dockertypes.ContainerJSON, error)
    CreateContainer(dockertypes.ContainerCreateConfig) (*dockertypes.ContainerCreateResponse, error)
    StartContainer(id string) error
    StopContainer(id string, timeout int) error
    RemoveContainer(id string, opts dockertypes.ContainerRemoveOptions) error
    InspectImage(image string) (*dockertypes.ImageInspect, error)
    ListImages(opts dockertypes.ImageListOptions) ([]dockertypes.Image, error)
    PullImage(image string, auth dockertypes.AuthConfig, opts dockertypes.ImagePullOptions) error
    RemoveImage(image string, opts dockertypes.ImageRemoveOptions) ([]dockertypes.ImageDelete, error)
    ImageHistory(id string) ([]dockertypes.ImageHistory, error)
    Logs(string, dockertypes.ContainerLogsOptions, StreamOptions) error
    Version() (*dockertypes.Version, error)
    Info() (*dockertypes.Info, error)
    CreateExec(string, dockertypes.ExecConfig) (*dockertypes.ContainerExecCreateResponse, error)
    StartExec(string, dockertypes.ExecStartCheck, StreamOptions) error
    InspectExec(id string) (*dockertypes.ContainerExecInspect, error)
    AttachToContainer(string, dockertypes.ContainerAttachOptions, StreamOptions) error
}

而我们最终初始化返回了结构体kubeDockerClient,所以DockerInterface接口的实现，我们可以回到kubeDockerClient结构体所在文件pkg/kubelet/dockertools/kube_docker_client.go查看接口实现。

DockeClient接口分析

源码目录： pkg/kubelet/dockertools/kube_docker_client.go
实现的接口如下：

可以看到kubeDockerClient结构体实现了所有的DockerInterface接口。
这些接口其实是对docker的操作接口进行了封装，下面取一个接口进行分析：

func (d *kubeDockerClient) ListContainers(options dockertypes.ContainerListOptions) ([]dockertypes.Container, error) {
    ctx, cancel := d.getTimeoutContext()
    defer cancel()
    containers, err := d.client.ContainerList(ctx, options)
    if ctxErr := contextError(ctx); ctxErr != nil {
        return nil, ctxErr
    }
    if err != nil {
        return nil, err
    }
    return containers, nil
}

该ListContainers()接口的关键就是调用了d.client.ContainerList(ctx, options).
所以关键对象还是client，继续回到上面讲初始化时介绍到的Client结构体。
Client结构所在文件： vendor/github.com/docker/engine-api/client/client.go
Client package结构：

操作docker API的接口都封装在这些文件中，有空可以深入了解下，这里就不一一介绍了，我们继续回到d.client.ContainerList(ctx, options),实现如下：

func (cli *Client) ContainerList(ctx context.Context, options types.ContainerListOptions) ([]types.Container, error) {
    query := url.Values{}

    if options.All {
        query.Set("all", "1")
    }

    if options.Limit != -1 {
        query.Set("limit", strconv.Itoa(options.Limit))
    }

    if options.Since != "" {
        query.Set("since", options.Since)
    }

    if options.Before != "" {
        query.Set("before", options.Before)
    }

    if options.Size {
        query.Set("size", "1")
    }

    if options.Filter.Len() > 0 {
        filterJSON, err := filters.ToParamWithVersion(cli.version, options.Filter)

        if err != nil {
            return nil, err
        }

        query.Set("filters", filterJSON)
    }

    resp, err := cli.get(ctx, "/containers/json", query, nil)
    if err != nil {
        return nil, err
    }

    var containers []types.Container
    err = json.NewDecoder(resp.body).Decode(&containers)
    ensureReaderClosed(resp)
    return containers, err
}

前面都是一些参数初始化，其实就是构建一个GET请求，然后调用cli.get()，该get就是一个httpRequest:

func (cli *Client) get(ctx context.Context, path string, query url.Values, headers map[string][]string) (*serverResponse, error) {
    return cli.sendRequest(ctx, "GET", path, query, nil, headers)
}

func (cli *Client) sendRequest(ctx context.Context, method, path string, query url.Values, obj interface{}, headers map[string][]string) (*serverResponse, error) {
    var body io.Reader

    if obj != nil {
        var err error
        body, err = encodeData(obj)
        if err != nil {
            return nil, err
        }
        if headers == nil {
            headers = make(map[string][]string)
        }
        headers["Content-Type"] = []string{"application/json"}
    }

    return cli.sendClientRequest(ctx, method, path, query, body, headers)
}

func (cli *Client) sendClientRequest(ctx context.Context, method, path string, query url.Values, body io.Reader, headers map[string][]string) (*serverResponse, error) {
    serverResp := &serverResponse{
        body:       nil,
        statusCode: -1,
    }

...
    req, err := cli.newRequest(method, path, query, body, headers)
    if cli.proto == "unix" || cli.proto == "npipe" {
        // For local communications, it doesn't matter what the host is. We just
        // need a valid and meaningful host name. (See #189)
        req.Host = "docker"
    }
    req.URL.Host = cli.addr
    req.URL.Scheme = cli.transport.Scheme()

    if expectedPayload && req.Header.Get("Content-Type") == "" {
        req.Header.Set("Content-Type", "text/plain")
    }

    resp, err := cancellable.Do(ctx, cli.transport, req)
    if resp != nil {
        serverResp.statusCode = resp.StatusCode
    }

...

    if serverResp.statusCode < 200 || serverResp.statusCode >= 400 {
        body, err := ioutil.ReadAll(resp.Body)
        if err != nil {
            return serverResp, err
        }
        if len(body) == 0 {
            return serverResp, fmt.Errorf("Error: request returned %s for API route and version %s, check if the server supports the requested API version", http.StatusText(serverResp.statusCode), req.URL)
        }
        return serverResp, fmt.Errorf("Error response from daemon: %s", bytes.TrimSpace(body))
    }

    serverResp.body = resp.Body
    serverResp.header = resp.Header
    return serverResp, nil
}

func Do(ctx context.Context, client transport.Sender, req *http.Request) (*http.Response, error) {
...
    result := make(chan responseAndError, 1)

    go func() {
        resp, err := client.Do(req)
        testHookDoReturned()
        result <- responseAndError{resp, err}
    }()

    var resp *http.Response

    select {
    case <-ctx.Done():
        testHookContextDoneBeforeHeaders()
        cancel()
        // Clean up after the goroutine calling client.Do:
        go func() {
            if r := <-result; r.resp != nil && r.resp.Body != nil {
                testHookDidBodyClose()
                r.resp.Body.Close()
            }
        }()
        return nil, ctx.Err()
    case r := <-result:
        var err error
        resp, err = r.resp, r.err
        if err != nil {
            return resp, err
        }
    }

...

    return resp, nil
}

上面列出了httpRequest的整个调用过程，最终调用client.Do(),该client对象需要回到之前的初始化过程中去，实际就是调用vemdor/github.com/docker/engine-api/client/client.go中的Client.transport，而该对象初始化时设置为apiTransport对象：

type apiTransport struct {
    *http.Client
    *tlsInfo
    transport *http.Transport
}

所以client.Do()实际就是调用http.Client.Do()。
OK,到此算是分析结束，具体的各个接口实现，还是需要花时间查看源码,但也都是大同小异。

学习源码的过程中，可以看到很多经典的实现，比如上面介绍的cancellable.Do()接口实现，golang非常推崇的"协程+channel"的方式，通过select case的方式循环等待协程处理的结果，确实很方便。