feat: Workflow job based ephemeral runner scaling (#721)

This add support for two upcoming enhancements on the GitHub side of self-hosted runners, ephemeral runners, and `workflow_jow` events. You can't use these yet.

**These features are not yet generally available to all GitHub users**. Please take this pull request as a preparation to make it available to actions-runner-controller users as soon as possible after GitHub released the necessary features on their end.

**Ephemeral runners**:

The former, ephemeral runners, is basically the reliable alternative to `--once`, which we've been using when you enabled `ephemeral: true` (default in actions-runner-controller).

`--once` has been suffering from a race issue #466. `--ephemeral` fixes that.

To enable ephemeral runners with `actions/runner`, you give `--ephemeral` to `config.sh`. This updated version of `actions-runner-controller` does it for you, by using `--ephemeral` instead of `--once` when you set `RUNNER_FEATURE_FLAG_EPHEMERAL=true`.

Please read the section `Ephemeral Runners` in the updated version of our README for more information.

Note that ephemeral runners is not released on GitHub yet. And `RUNNER_FEATURE_FLAG_EPHEMERAL=true` won't work at all until the feature gets released on GitHub. Stay tuned for an announcement from GitHub!

**`workflow_job` events**:

`workflow_job` is the additional webhook event that corresponds to each GitHub Actions workflow job run. It provides `actions-runner-controller` a solid foundation to improve our webhook-based autoscale.

Formerly, we've been exploiting webhook events like `check_run` for autoscaling. However, as none of our supported events has included `labels`, you had to configure an HRA to only match relevant `check_run` events. It wasn't trivial.

In contrast, a `workflow_job` event payload contains `labels` of runners requested. `actions-runner-controller` is able to automatically decide which HRA to scale by filtering the corresponding RunnerDeployment by `labels` included in the webhook payload. So all you need to use webhook-based autoscale will be to enable `workflow_job` on GitHub and expose actions-runner-controller's webhook server to the internet.

Note that the current implementation of `workflow_job` support works in two ways, increment, and decrement. An increment happens when the webhook server receives` workflow_job` of `queued` status. A decrement happens when it receives `workflow_job` of `completed` status. The latter is used to make scaling-down faster so that you waste money less than before. You still don't suffer from flapping, as a scale-down is still subject to `scaleDownDelaySecondsAfterScaleOut `.

Please read the section `Example 3: Scale on each `workflow_job` event` in the updated version of our README for more information on its usage.
This commit is contained in:
Yusuke Kuoka
2021-08-11 09:52:04 +09:00
committed by GitHub
parent d528d18211
commit fabead8c8e
9 changed files with 338 additions and 16 deletions

View File

@@ -183,6 +183,45 @@ func (autoscaler *HorizontalRunnerAutoscalerGitHubWebhook) Handle(w http.Respons
"action", e.GetAction(),
)
}
case *gogithub.WorkflowJobEvent:
if workflowJob := e.GetWorkflowJob(); workflowJob != nil {
log = log.WithValues(
"workflowJob.status", workflowJob.GetStatus(),
"workflowJob.labels", workflowJob.Labels,
"repository.name", e.Repo.GetName(),
"repository.owner.login", e.Repo.Owner.GetLogin(),
"repository.owner.type", e.Repo.Owner.GetType(),
"action", e.GetAction(),
)
}
labels := e.WorkflowJob.Labels
switch e.GetAction() {
case "queued", "completed":
target, err = autoscaler.getJobScaleUpTargetForRepoOrOrg(
context.TODO(),
log,
e.Repo.GetName(),
e.Repo.Owner.GetLogin(),
e.Repo.Owner.GetType(),
labels,
)
if target != nil {
if e.GetAction() == "queued" {
target.Amount = 1
} else if e.GetAction() == "completed" {
// A nagative amount is processed in the tryScale func as a scale-down request,
// that erasese the oldest CapacityReservation with the same amount.
// If the first CapacityReservation was with Replicas=1, this negative scale target erases that,
// so that the resulting desired replicas decreases by 1.
target.Amount = -1
}
}
default:
}
case *gogithub.PingEvent:
ok = true
@@ -227,7 +266,7 @@ func (autoscaler *HorizontalRunnerAutoscalerGitHubWebhook) Handle(w http.Respons
return
}
if err := autoscaler.tryScaleUp(context.TODO(), target); err != nil {
if err := autoscaler.tryScale(context.TODO(), target); err != nil {
log.Error(err, "could not scale up")
return
@@ -237,7 +276,7 @@ func (autoscaler *HorizontalRunnerAutoscalerGitHubWebhook) Handle(w http.Respons
w.WriteHeader(http.StatusOK)
msg := fmt.Sprintf("scaled %s by 1", target.Name)
msg := fmt.Sprintf("scaled %s by %d", target.Name, target.Amount)
autoscaler.Log.Info(msg)
@@ -394,7 +433,137 @@ func (autoscaler *HorizontalRunnerAutoscalerGitHubWebhook) getScaleUpTarget(ctx
return nil, nil
}
func (autoscaler *HorizontalRunnerAutoscalerGitHubWebhook) tryScaleUp(ctx context.Context, target *ScaleTarget) error {
func (autoscaler *HorizontalRunnerAutoscalerGitHubWebhook) getJobScaleUpTargetForRepoOrOrg(ctx context.Context, log logr.Logger, repo, owner, ownerType string, labels []string) (*ScaleTarget, error) {
repositoryRunnerKey := owner + "/" + repo
if target, err := autoscaler.getJobScaleTarget(ctx, repositoryRunnerKey, labels); err != nil {
log.Info("finding repository-wide runner", "repository", repositoryRunnerKey)
return nil, err
} else if target != nil {
log.Info("job scale up target is repository-wide runners", "repository", repo)
return target, nil
}
if ownerType == "User" {
log.V(1).Info("no repository runner found", "organization", owner)
return nil, nil
}
if target, err := autoscaler.getJobScaleTarget(ctx, owner, labels); err != nil {
log.Info("finding organizational runner", "organization", owner)
return nil, err
} else if target != nil {
log.Info("job scale up target is organizational runners", "organization", owner)
return target, nil
} else {
log.V(1).Info("no repository runner or organizational runner found",
"repository", repositoryRunnerKey,
"organization", owner,
)
}
return nil, nil
}
func (autoscaler *HorizontalRunnerAutoscalerGitHubWebhook) getJobScaleTarget(ctx context.Context, name string, labels []string) (*ScaleTarget, error) {
hras, err := autoscaler.findHRAsByKey(ctx, name)
if err != nil {
return nil, err
}
autoscaler.Log.V(1).Info(fmt.Sprintf("Found %d HRAs by key", len(hras)), "key", name)
HRA:
for _, hra := range hras {
if !hra.ObjectMeta.DeletionTimestamp.IsZero() {
continue
}
if len(hra.Spec.ScaleUpTriggers) > 1 {
autoscaler.Log.V(1).Info("Skipping this HRA as it has too many ScaleUpTriggers to be used in workflow_job based scaling", "hra", hra.Name)
continue
}
var duration metav1.Duration
if len(hra.Spec.ScaleUpTriggers) > 0 {
duration = hra.Spec.ScaleUpTriggers[0].Duration
}
if duration.Duration <= 0 {
// Try to release the reserved capacity after at least 10 minutes by default,
// we won't end up in the reserved capacity remained forever in case GitHub somehow stopped sending us "completed" workflow_job events.
// GitHub usually send us those but nothing is 100% guaranteed, e.g. in case of something went wrong on GitHub :)
// Probably we'd better make this configurable via custom resources in the future?
duration.Duration = 10 * time.Minute
}
switch hra.Spec.ScaleTargetRef.Kind {
case "RunnerSet":
var rs v1alpha1.RunnerSet
if err := autoscaler.Client.Get(context.Background(), types.NamespacedName{Namespace: hra.Namespace, Name: hra.Spec.ScaleTargetRef.Name}, &rs); err != nil {
return nil, err
}
if len(labels) == 1 && labels[0] == "self-hosted" {
return &ScaleTarget{HorizontalRunnerAutoscaler: hra, ScaleUpTrigger: v1alpha1.ScaleUpTrigger{Duration: duration}}, nil
}
// Ensure that the RunnerSet-managed runners have all the labels requested by the workflow_job.
for _, l := range labels {
var matched bool
for _, l2 := range rs.Spec.Labels {
if l == l2 {
matched = true
break
}
}
if !matched {
continue HRA
}
}
return &ScaleTarget{HorizontalRunnerAutoscaler: hra, ScaleUpTrigger: v1alpha1.ScaleUpTrigger{Duration: duration}}, nil
case "RunnerDeployment", "":
var rd v1alpha1.RunnerDeployment
if err := autoscaler.Client.Get(context.Background(), types.NamespacedName{Namespace: hra.Namespace, Name: hra.Spec.ScaleTargetRef.Name}, &rd); err != nil {
return nil, err
}
if len(labels) == 1 && labels[0] == "self-hosted" {
return &ScaleTarget{HorizontalRunnerAutoscaler: hra, ScaleUpTrigger: v1alpha1.ScaleUpTrigger{Duration: duration}}, nil
}
// Ensure that the RunnerDeployment-managed runners have all the labels requested by the workflow_job.
for _, l := range labels {
var matched bool
for _, l2 := range rd.Spec.Template.Labels {
if l == l2 {
matched = true
break
}
}
if !matched {
continue HRA
}
}
return &ScaleTarget{HorizontalRunnerAutoscaler: hra, ScaleUpTrigger: v1alpha1.ScaleUpTrigger{Duration: duration}}, nil
default:
return nil, fmt.Errorf("unsupported scaleTargetRef.kind: %v", hra.Spec.ScaleTargetRef.Kind)
}
}
return nil, nil
}
func (autoscaler *HorizontalRunnerAutoscalerGitHubWebhook) tryScale(ctx context.Context, target *ScaleTarget) error {
if target == nil {
return nil
}
@@ -403,16 +572,38 @@ func (autoscaler *HorizontalRunnerAutoscalerGitHubWebhook) tryScaleUp(ctx contex
amount := 1
if target.ScaleUpTrigger.Amount > 0 {
if target.ScaleUpTrigger.Amount != 0 {
amount = target.ScaleUpTrigger.Amount
}
capacityReservations := getValidCapacityReservations(copy)
copy.Spec.CapacityReservations = append(capacityReservations, v1alpha1.CapacityReservation{
ExpirationTime: metav1.Time{Time: time.Now().Add(target.ScaleUpTrigger.Duration.Duration)},
Replicas: amount,
})
if amount > 0 {
copy.Spec.CapacityReservations = append(capacityReservations, v1alpha1.CapacityReservation{
ExpirationTime: metav1.Time{Time: time.Now().Add(target.ScaleUpTrigger.Duration.Duration)},
Replicas: amount,
})
} else if amount < 0 {
var reservations []v1alpha1.CapacityReservation
var found bool
for _, r := range capacityReservations {
if !found && r.Replicas+amount == 0 {
found = true
} else {
reservations = append(reservations, r)
}
}
copy.Spec.CapacityReservations = reservations
}
autoscaler.Log.Info(
"Patching hra for capacityReservations update",
"before", target.HorizontalRunnerAutoscaler.Spec.CapacityReservations,
"after", copy.Spec.CapacityReservations,
)
if err := autoscaler.Client.Patch(ctx, copy, client.MergeFrom(&target.HorizontalRunnerAutoscaler)); err != nil {
return fmt.Errorf("patching horizontalrunnerautoscaler to add capacity reservation: %w", err)
@@ -450,13 +641,26 @@ func (autoscaler *HorizontalRunnerAutoscalerGitHubWebhook) SetupWithManager(mgr
return nil
}
var rd v1alpha1.RunnerDeployment
switch hra.Spec.ScaleTargetRef.Kind {
case "", "RunnerDeployment":
var rd v1alpha1.RunnerDeployment
if err := autoscaler.Client.Get(context.Background(), types.NamespacedName{Namespace: hra.Namespace, Name: hra.Spec.ScaleTargetRef.Name}, &rd); err != nil {
return nil
if err := autoscaler.Client.Get(context.Background(), types.NamespacedName{Namespace: hra.Namespace, Name: hra.Spec.ScaleTargetRef.Name}, &rd); err != nil {
return nil
}
return []string{rd.Spec.Template.Spec.Repository, rd.Spec.Template.Spec.Organization}
case "RunnerSet":
var rs v1alpha1.RunnerSet
if err := autoscaler.Client.Get(context.Background(), types.NamespacedName{Namespace: hra.Namespace, Name: hra.Spec.ScaleTargetRef.Name}, &rs); err != nil {
return nil
}
return []string{rs.Spec.Repository, rs.Spec.Organization}
}
return []string{rd.Spec.Template.Spec.Repository, rd.Spec.Template.Spec.Organization}
return nil
}); err != nil {
return err
}