github/kubernetes
4
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge pull request #122643 from soltysh/never_terminate

Browse Source 
Add a new neverTerminate job behavior just for upgrade
This commit is contained in:
Kubernetes Prow Robot
2024-01-08 14:54:13 +01:00
committed by GitHub
parent 18608cc56b f8abe71238
commit 4142dda1e9
2 changed files with 28 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
test/e2e/framework/job/fixtures.go
View File

@@ -92,6 +92,12 @@ func NewTestJobOnNode(behavior, name string, rPol v1.RestartPolicy, parallelism,
}
}
switch behavior {
case "neverTerminate":
// this job is being used in an upgrade job see test/e2e/upgrades/apps/job.go
// it should never be optimized, as it always has to restart during an upgrade
// and continue running
job.Spec.Template.Spec.Containers[0].Command = []string{"sleep", "1000000"}
job.Spec.Template.Spec.TerminationGracePeriodSeconds = ptr.To(int64(1))
case "notTerminate":
job.Spec.Template.Spec.Containers[0].Image = imageutils.GetPauseImageName()
case "fail":

25
test/e2e/upgrades/apps/job.go
View File

@@ -47,7 +47,7 @@ func (t *JobUpgradeTest) Setup(ctx context.Context, f *framework.Framework) {
t.namespace = f.Namespace.Name
ginkgo.By("Creating a job")
t.job = e2ejob.NewTestJob("notTerminate", "foo", v1.RestartPolicyOnFailure, 2, 2, nil, 6)
t.job = e2ejob.NewTestJob("neverTerminate", "foo", v1.RestartPolicyOnFailure, 2, 2, nil, 6)
job, err := e2ejob.CreateJob(ctx, f.ClientSet, t.namespace, t.job)
t.job = job
framework.ExpectNoError(err)
@@ -60,8 +60,11 @@ func (t *JobUpgradeTest) Setup(ctx context.Context, f *framework.Framework) {
// Test verifies that the Jobs Pods are running after the an upgrade
func (t *JobUpgradeTest) Test(ctx context.Context, f *framework.Framework, done <-chan struct{}, upgrade upgrades.UpgradeType) {
<-done
ginkgo.By("Ensuring job is running")
err := ensureJobRunning(ctx, f.ClientSet, t.namespace, t.job.Name)
framework.ExpectNoError(err)
ginkgo.By("Ensuring active pods == parallelism")
err := ensureAllJobPodsRunning(ctx, f.ClientSet, t.namespace, t.job.Name, 2)
err = ensureAllJobPodsRunning(ctx, f.ClientSet, t.namespace, t.job.Name, 2)
framework.ExpectNoError(err)
}
@@ -70,7 +73,7 @@ func (t *JobUpgradeTest) Teardown(ctx context.Context, f *framework.Framework) {
// rely on the namespace deletion to clean up everything
}
// ensureAllJobPodsRunning uses c to check in the Job named jobName in ns
// ensureAllJobPodsRunning uses c to check if the Job named jobName in ns
// is running, returning an error if the expected parallelism is not
// satisfied.
func ensureAllJobPodsRunning(ctx context.Context, c clientset.Interface, ns, jobName string, parallelism int32) error {
@@ -93,3 +96,19 @@ func ensureAllJobPodsRunning(ctx context.Context, c clientset.Interface, ns, job
}
return nil
}
// ensureJobRunning uses c to check if the Job named jobName in ns is running,
// (not completed, nor failed, nor suspended) returning an error if it can't
// read the job or when it's not runnig
func ensureJobRunning(ctx context.Context, c clientset.Interface, ns, jobName string) error {
job, err := e2ejob.GetJob(ctx, c, ns, jobName)
if err != nil {
return err
}
for _, c := range job.Status.Conditions {
if (c.Type == batchv1.JobComplete || c.Type == batchv1.JobFailed || c.Type == batchv1.JobSuspended) && c.Status == v1.ConditionTrue {
return fmt.Errorf("job is not running %#v", job)
}
}
return nil
}