What is the method called by the containerd source code interface
This article mainly explains "what is the method of calling containerd source code interface". Interested friends may wish to have a look at it. The method introduced in this paper is simple, fast and practical. Let's let the editor take you to learn "what is the method of calling the containerd source interface?"
Source code API call details call containerd-api from ctr
# checkpoint (for snapshots, the function of docker is not perfect) | ctr cmd | containerd-api | |-| | list | / types.API/ListCheckpoint | | create | / types.API/CreateCheckpoint | | delete | / types.API/DeleteCheckpoint |
# containers | ctr cmd | containerd-api | |-| | list, state | / types.API/State | | pause, resume, update | / types.API/UpdateContainer | | create | / types.API/CreateContainer | | stats | / types.API/Stats | | watch | / types.API/State, / types.API/Events | | exec | / types.API/Events, / types.API/AddProcess / types.API/UpdateProcess | | kill | / types.API/Signal | | start | / types.API/Events, / types.API/CreateContainer, / types.API/UpdateProcess | | update | / types.API/UpdateContainer | # events / types.API/Events # state / types.API/State # version / types.API/GetServerVersion-- return result
Task processing from containerd-api to supervisor
Note: API--server.go-> daemon-supervisor.go (handleTask func) # checkpoint | containerd-api | supervisor | |-| | / types.API/ListCheckpoint (supervisor.GetContainersTask) | getContainers | | / types.API/CreateCheckpoint | createCheckpoint | | / types.API/DeleteCheckpoint | deleteCheckpoint |
# containers | containerd-api | supervisor | |-| | / types.API/State / types.API/Stats (supervisor.GetContainersTask) | getContainers | | / types.API/UpdateContainer (supervisor.UpdateTask) | updateContainer | | / types.API/CreateContainer (supervisor.StartTask) | start | | / types.API/Events | Events | | / types.API/AddProcess | addProcess | / types.API/UpdateProcess | updateProcess | | / types.API/Signal | signal |
From supervisor to runtime (runC)
# checkpoint | supervisor | runtime | |-| | getContainers |-| | createCheckpoint | (runtime) CheckPoint-- > exec.Command (c.runtimemilitary arg....) | | deleteCheckpoint | (runtime) DeleteCheckpoint | # containers | supervisor | runtime |-| getContainers |-| updateContainer | (runtime) Resume Pause UpdateResources-- > exec.Command (c.runtime) Arg....) | | start | (runtime supervisor/worker.go) Start-- > exec.Command (c.shimforce c.idrec.bundlerec.runtime) | | addProcess | (runtime) exec-- > exec.Command (c.shimjournal c.bundlerec.runtime) | | updateProcess |-| signal |-|
# # take createContainer as an example to read the code # deamon starts listening to tasks and startTasks process # enter the main.go main method and call the daemon method
App.Action = func (context * cli.Context) {if err: = daemon (context); err! = nil {logrus.Fatal (err)}}
# enter main.go daemon method
For I: = 0; I < 10; iTunes + {wg.Add (1) w: = supervisor.NewWorker (sv, wg) go w.Start ()} if err: = sv.Start (); err! = nil {return err}
# initialize supervisor/worker.go NewWorker and start listening startTask and processing
Func NewWorker (s * Supervisor, wg * sync.WaitGroup) Worker {return & worker {s: s, wg: wg,}} func (w * worker) Start () {defer w.wg.Done () for t: = range w.s.startTasks {started: = time.Now () process, err: = t.Container.Start (t.CheckpointPath, runtime.NewStdio (t.Stdin) T.Stdout, t.Stderr) if err! = nil {logrus.WithFields (logrus.Fields {"error": err, "id": t.Container.ID (),}) .Error ("containerd: start container") t.Err