diff --git a/test/e2e/basic/sysctl_d_override.go b/test/e2e/basic/sysctl_d_override.go
index edfb7215ed..00f3a8cb71 100644
--- a/test/e2e/basic/sysctl_d_override.go
+++ b/test/e2e/basic/sysctl_d_override.go
@@ -59,9 +59,11 @@ var _ = ginkgo.Describe("[basic][sysctl_d_override] Node Tuning Operator /etc/sy
 
 			ginkgo.By(fmt.Sprintf("writing %s override file on the host with %s=%s", sysctlFile, sysctlVar, sysctlValSet))
 			_, _, err = util.ExecAndLogCommand("oc", "exec", "-n", ntoconfig.OperatorNamespace(), pod.Name, "--", "sh", "-c",
-				fmt.Sprintf("echo %s=%s > %s", sysctlVar, sysctlValSet, sysctlFile))
+				fmt.Sprintf("echo %s=%s > %s; sync %s", sysctlVar, sysctlValSet, sysctlFile, sysctlFile))
 			gomega.Expect(err).NotTo(gomega.HaveOccurred())
 
+			util.ExecAndLogCommand("oc", "rsh", "-n", ntoconfig.OperatorNamespace(), pod.Name, "cat", sysctlFile)
+
 			ginkgo.By(fmt.Sprintf("deleting Pod %s", pod.Name))
 			_, _, err = util.ExecAndLogCommand("oc", "delete", "-n", ntoconfig.OperatorNamespace(), "pod", pod.Name, "--wait")
 			gomega.Expect(err).NotTo(gomega.HaveOccurred())
diff --git a/test/e2e/reboots/stalld.go b/test/e2e/reboots/stalld.go
index 40be97f274..957bd160f8 100644
--- a/test/e2e/reboots/stalld.go
+++ b/test/e2e/reboots/stalld.go
@@ -88,6 +88,7 @@ var _ = ginkgo.Describe("[reboots][stalld] Node Tuning Operator installing syste
 
 			ginkgo.By(fmt.Sprintf("checking the stalld daemon is not running on node %s", node.Name))
 			out, err := util.ExecCmdInPod(pod, "pidof", "stalld")
+			gomega.Expect(out).To(gomega.Equal(""))
 			gomega.Expect(err).To(gomega.HaveOccurred()) // pidof exits 1 when there is no running process found
 
 			ginkgo.By(fmt.Sprintf("applying custom realtime profile %s with stalld service", profileStalldOn))
@@ -99,28 +100,6 @@ var _ = ginkgo.Describe("[reboots][stalld] Node Tuning Operator installing syste
 			util.Logf(fmt.Sprintf("stalld process running on node %s with PID %s", node.Name, out))
 			gomega.Expect(err).NotTo(gomega.HaveOccurred())
 
-			// We need to wait for TuneD daemon to start and apply its profile, otherwise it might race
-			// to start 'stalld' after we stop it via systemctl below.
-			waitForTuneD()
-
-			ginkgo.By(fmt.Sprintf("stopping the stalld daemon on node %s", node.Name))
-			out, err = util.WaitForCmdInPod(pollInterval, waitDuration, pod, "chroot", "/host", "systemctl", "stop", "stalld")
-			util.ExecAndLogCommand("oc", "rsh", "-n", ntoconfig.OperatorNamespace(), pod.Name, "chroot", "/host", "systemctl", "status", "stalld", "-l", "--no-pager")
-			gomega.Expect(err).NotTo(gomega.HaveOccurred())
-
-			ginkgo.By(fmt.Sprintf("checking the stalld daemon is not running on node %s", node.Name))
-			_, err = util.ExecCmdInPod(pod, "pidof", "stalld")
-			gomega.Expect(err).To(gomega.HaveOccurred()) // pidof exits 1 when there is no running process found
-
-			ginkgo.By(fmt.Sprintf("rebooting node %s with stalld daemon enabled", node.Name))
-			util.ExecCmdInPod(pod, "chroot", "/host", "reboot")
-			// Ignore errors; we can get "exit status 143", i.e. SIGTERM caused by the reboot
-
-			// Wait for the host to reboot and the TuneD [service] plug-in to start/enable stalld service.
-			ginkgo.By(fmt.Sprintf("checking the stalld daemon is running on node %s", node.Name))
-			out, err = util.WaitForCmdInPod(pollInterval, 20*time.Minute, pod, "pidof", "stalld")
-			gomega.Expect(err).NotTo(gomega.HaveOccurred())
-
 			// Node label needs to be removed first, and we also need to wait for the worker pool to complete the update;
 			// otherwise worker-rt MachineConfigPool deletion would cause Degraded state of the worker pool.
 			// see rhbz#1816239