Removed .service file for healthchecks

when a container with healthchecks exits due to stopping or failure, we
need the cleanup process to remove both the timer file and the service
file.

Bz#:2024229

Signed-off-by: Brent Baude <bbaude@redhat.com>

3 files changed, 34 insertions, 10 deletions

diff --git a/libpod/container_internal.go b/libpod/container_internal.go
index a68de3173..7df82eb18 100644
--- a/libpod/container_internal.go
+++ b/libpod/container_internal.go
@@ -1877,7 +1877,7 @@ func (c *Container) cleanupStorage() error {
 	return cleanupErr
 }
 
-// Unmount the a container and free its resources
+// Unmount the container and free its resources
 func (c *Container) cleanup(ctx context.Context) error {
 	var lastError error
 
@@ -1885,7 +1885,7 @@ func (c *Container) cleanup(ctx context.Context) error {
 
 	// Remove healthcheck unit/timer file if it execs
 	if c.config.HealthCheckConfig != nil {
-		if err := c.removeTimer(); err != nil {
+		if err := c.removeTransientFiles(ctx); err != nil {
 			logrus.Errorf("Removing timer for container %s healthcheck: %v", c.ID(), err)
 		}
 	}
diff --git a/libpod/healthcheck_linux.go b/libpod/healthcheck_linux.go
index e08214809..2c19e0a61 100644
--- a/libpod/healthcheck_linux.go
+++ b/libpod/healthcheck_linux.go
@@ -1,6 +1,7 @@
 package libpod
 
 import (
+	"context"
 	"fmt"
 	"os"
 	"os/exec"
@@ -59,9 +60,9 @@ func (c *Container) startTimer() error {
 	return err
 }
 
-// removeTimer removes the systemd timer and unit files
+// removeTransientFiles removes the systemd timer and unit files
 // for the container
-func (c *Container) removeTimer() error {
+func (c *Container) removeTransientFiles(ctx context.Context) error {
 	if c.disableHealthCheckSystemd() {
 		return nil
 	}
@@ -71,12 +72,19 @@ func (c *Container) removeTimer() error {
 	}
 	defer conn.Close()
 	timerFile := fmt.Sprintf("%s.timer", c.ID())
-	_, err = conn.StopUnit(timerFile, "fail", nil)
-
-	// We want to ignore errors where the timer unit has already been removed. The error
-	// return is generic so we have to check against the string in the error
-	if err != nil && strings.HasSuffix(err.Error(), ".timer not loaded.") {
-		return nil
+	serviceFile := fmt.Sprintf("%s.service", c.ID())
+	// We want to ignore errors where the timer unit and/or service unit has already
+	// been removed. The error return is generic so we have to check against the
+	// string in the error
+	if _, err = conn.StopUnitContext(ctx, serviceFile, "fail", nil); err != nil {
+		if !strings.HasSuffix(err.Error(), ".service not loaded.") {
+			return errors.Wrapf(err, "unable to remove service file")
+		}
+	}
+	if _, err = conn.StopUnitContext(ctx, timerFile, "fail", nil); err != nil {
+		if strings.HasSuffix(err.Error(), ".timer not loaded.") {
+			return nil
+		}
 	}
 	return err
 }
diff --git a/test/e2e/healthcheck_run_test.go b/test/e2e/healthcheck_run_test.go
index 6e4dc5bbf..c2084a6fd 100644
--- a/test/e2e/healthcheck_run_test.go
+++ b/test/e2e/healthcheck_run_test.go
@@ -242,4 +242,20 @@ var _ = Describe("Podman healthcheck run", func() {
 		Expect(ps.OutputToStringArray()).To(HaveLen(2))
 		Expect(ps.OutputToString()).To(ContainSubstring("hc"))
 	})
+
+	It("stopping and then starting a container with healthcheck cmd", func() {
+		session := podmanTest.Podman([]string{"run", "-dt", "--name", "hc", "--health-cmd", "[\"ls\", \"/foo\"]", ALPINE, "top"})
+		session.WaitWithDefaultTimeout()
+		Expect(session).Should(Exit(0))
+
+		stop := podmanTest.Podman([]string{"stop", "-t0", "hc"})
+		stop.WaitWithDefaultTimeout()
+		Expect(stop).Should(Exit(0))
+
+		startAgain := podmanTest.Podman([]string{"start", "hc"})
+		startAgain.WaitWithDefaultTimeout()
+		Expect(startAgain).Should(Exit(0))
+		Expect(startAgain.OutputToString()).To(Equal("hc"))
+		Expect(startAgain.ErrorToString()).To(Equal(""))
+	})
 })