Cleanup kube play workloads if error happens

If an error happening while playing a kube yaml,
clean up any pods, containers, and volumes that might
have been created before the error was hit.
This improves the user experience for when they go to
re-run the same yaml with their fixes and podman doesn't
complain about any existing workloads from the previously
failed run.

Suppress the clean up output when clean up happens after an
error as the user doesn't need to see or know about that.

Signed-off-by: Urvashi Mohnani <umohnani@redhat.com>

test/e2e/play_kube_test.go

@@ -4406,4 +4406,40 @@ cgroups="disabled"`), 0644)
 		kube.WaitWithDefaultTimeout()
 		Expect(kube).Should(Exit(0))
 	})
+
+	It("podman kube play invalid yaml should clean up pod that was created before failure", func() {
+		podTemplate := `---
+apiVersion: v1
+kind: Pod
+metadata:
+	creationTimestamp: "2022-08-02T04:05:53Z"
+	labels:
+	app: vol-test-3-pod
+	name: vol-test-3
+spec:
+	containers:
+	- command:
+	- sleep
+	- "1000"
+	image: non-existing-image
+	name: vol-test-3
+	securityContext:
+		capabilities:
+		drop:
+		- CAP_MKNOD
+		- CAP_NET_RAW
+		- CAP_AUDIT_WRITE
+`
+
+		// the image is incorrect so the kube play will fail, but it will clean up the pod that was created for it before the failure happened
+		kube := podmanTest.Podman([]string{"kube", "play", podTemplate})
+		kube.WaitWithDefaultTimeout()
+		Expect(kube).To(ExitWithError())
+
+		ps := podmanTest.Podman([]string{"pod", "ps", "-q"})
+		ps.WaitWithDefaultTimeout()
+		Expect(ps).Should(Exit(0))
+		Expect(ps.OutputToStringArray()).To(HaveLen(0))
+	})
+
 })