Restart stopped tasks

This patch reports stopped tasks as having zero scale, which
means the gateway will send a "scale up" request, the same
way as it does for paused containers, or those which have
no task due to a reboot of the machine.

The scale up logic will now delete the stopped task and
recreate the task.

Tested with nodeinfo and figlet on a Dell XPS with
Ubuntu 16.04. The scaling logic has been re-written, but
re-tested by manually pausing and manually removing
the task of a container.

Signed-off-by: Alex Ellis (OpenFaaS Ltd) <alexellis2@gmail.com>
This commit is contained in:
Alex Ellis (OpenFaaS Ltd)
2020-09-19 21:07:43 +01:00
committed by Alex Ellis
parent 87f49b0289
commit 40829bbf88
4 changed files with 67 additions and 40 deletions

View File

@ -11,6 +11,7 @@ import (
"github.com/containerd/containerd"
"github.com/containerd/containerd/namespaces"
gocni "github.com/containerd/go-cni"
"github.com/openfaas/faas-provider/types"
faasd "github.com/openfaas/faasd/pkg"
)
@ -58,46 +59,70 @@ func MakeReplicaUpdateHandler(client *containerd.Client, cni gocni.CNI) func(w h
return
}
taskExists := true
var taskExists bool
var taskStatus *containerd.Status
task, taskErr := ctr.Task(ctx, nil)
if taskErr != nil {
msg := fmt.Sprintf("cannot load task for service %s, error: %s", name, taskErr)
log.Printf("[Scale] %s\n", msg)
taskExists = false
} else {
taskExists = true
status, statusErr := task.Status(ctx)
if statusErr != nil {
msg := fmt.Sprintf("cannot load task status for %s, error: %s", name, statusErr)
log.Printf("[Scale] %s\n", msg)
http.Error(w, msg, http.StatusInternalServerError)
return
} else {
taskStatus = &status
}
}
if req.Replicas > 0 {
if taskExists {
if status, statusErr := task.Status(ctx); statusErr == nil {
if status.Status == containerd.Paused {
if resumeErr := task.Resume(ctx); resumeErr != nil {
log.Printf("[Scale] error resuming task %s, error: %s\n", name, resumeErr)
http.Error(w, resumeErr.Error(), http.StatusBadRequest)
}
}
}
} else {
deployErr := createTask(ctx, client, ctr, cni)
if deployErr != nil {
log.Printf("[Scale] error deploying %s, error: %s\n", name, deployErr)
http.Error(w, deployErr.Error(), http.StatusBadRequest)
createNewTask := false
// Scale to zero
if req.Replicas == 0 {
// If a task is running, pause it
if taskExists && taskStatus.Status == containerd.Running {
if pauseErr := task.Pause(ctx); pauseErr != nil {
wrappedPauseErr := fmt.Errorf("error pausing task %s, error: %s", name, pauseErr)
log.Printf("[Scale] %s\n", wrappedPauseErr.Error())
http.Error(w, wrappedPauseErr.Error(), http.StatusNotFound)
return
}
return
}
} else {
if taskExists {
if status, statusErr := task.Status(ctx); statusErr == nil {
if status.Status == containerd.Running {
if pauseErr := task.Pause(ctx); pauseErr != nil {
log.Printf("[Scale] error pausing task %s, error: %s\n", name, pauseErr)
http.Error(w, pauseErr.Error(), http.StatusBadRequest)
}
}
}
}
}
}
if taskExists {
if taskStatus != nil {
if taskStatus.Status == containerd.Paused {
if resumeErr := task.Resume(ctx); resumeErr != nil {
log.Printf("[Scale] error resuming task %s, error: %s\n", name, resumeErr)
http.Error(w, resumeErr.Error(), http.StatusBadRequest)
}
} else if taskStatus.Status == containerd.Stopped {
// Stopped tasks cannot be restarted, must be removed, and created again
if _, delErr := task.Delete(ctx); delErr != nil {
log.Printf("[Scale] error deleting stopped task %s, error: %s\n", name, delErr)
http.Error(w, delErr.Error(), http.StatusBadRequest)
}
createNewTask = true
}
}
} else {
createNewTask = true
}
if createNewTask {
deployErr := createTask(ctx, client, ctr, cni)
if deployErr != nil {
log.Printf("[Scale] error deploying %s, error: %s\n", name, deployErr)
http.Error(w, deployErr.Error(), http.StatusBadRequest)
return
}
}
}
}