fix: add missing service container health check (#2354)

* fix: Implement missing health ceck for Services

* Add test case

* linter doesn't support min builtin and fix check

---------

Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

pkg/container/container_types.go

@@ -54,6 +54,7 @@ type Container interface {
 	Remove() common.Executor
 	Close() common.Executor
 	ReplaceLogWriter(io.Writer, io.Writer) (io.Writer, io.Writer)
+	GetHealth(ctx context.Context) ContainerHealth
 }
 
 // NewDockerBuildExecutorInput the input for the NewDockerBuildExecutor function
@@ -73,3 +74,11 @@ type NewDockerPullExecutorInput struct {
 	Username  string
 	Password  string
 }
+
+type ContainerHealth int
+
+const (
+	ContainerHealthStarting ContainerHealth = iota
+	ContainerHealthHealthy
+	ContainerHealthUnHealthy
+)

pkg/container/docker_run.go

@@ -169,6 +169,30 @@ func (cr *containerReference) Remove() common.Executor {
 	).IfNot(common.Dryrun)
 }
 
+func (cr *containerReference) GetHealth(ctx context.Context) ContainerHealth {
+	resp, err := cr.cli.ContainerInspect(ctx, cr.id)
+	logger := common.Logger(ctx)
+	if err != nil {
+		logger.Errorf("failed to query container health %s", err)
+		return ContainerHealthUnHealthy
+	}
+	if resp.Config == nil || resp.Config.Healthcheck == nil || resp.State == nil || resp.State.Health == nil || len(resp.Config.Healthcheck.Test) == 1 && strings.EqualFold(resp.Config.Healthcheck.Test[0], "NONE") {
+		logger.Debugf("no container health check defined")
+		return ContainerHealthHealthy
+	}
+
+	logger.Infof("container health of %s (%s) is %s", cr.id, resp.Config.Image, resp.State.Health.Status)
+	switch resp.State.Health.Status {
+	case "starting":
+		return ContainerHealthStarting
+	case "healthy":
+		return ContainerHealthHealthy
+	case "unhealthy":
+		return ContainerHealthUnHealthy
+	}
+	return ContainerHealthUnHealthy
+}
+
 func (cr *containerReference) ReplaceLogWriter(stdout io.Writer, stderr io.Writer) (io.Writer, io.Writer) {
 	out := cr.input.Stdout
 	err := cr.input.Stderr

pkg/container/host_environment.go

@@ -452,6 +452,10 @@ func (e *HostEnvironment) GetRunnerContext(_ context.Context) map[string]interfa
 	}
 }
 
+func (e *HostEnvironment) GetHealth(ctx context.Context) ContainerHealth {
+	return ContainerHealthHealthy
+}
+
 func (e *HostEnvironment) ReplaceLogWriter(stdout io.Writer, _ io.Writer) (io.Writer, io.Writer) {
 	org := e.StdOut
 	e.StdOut = stdout

pkg/runner/run_context.go

@@ -17,6 +17,7 @@ import (
 	"runtime"
 	"strconv"
 	"strings"
+	"time"
 
 	"github.com/docker/go-connections/nat"
 	"github.com/nektos/act/pkg/common"
@@ -420,6 +421,7 @@ func (rc *RunContext) startJobContainer() common.Executor {
 				Mode: 0o666,
 				Body: "",
 			}),
+			rc.waitForServiceContainers(),
 		)(ctx)
 	}
 }
@@ -518,6 +520,40 @@ func (rc *RunContext) startServiceContainers(_ string) common.Executor {
 	}
 }
 
+func (rc *RunContext) waitForServiceContainer(c container.ExecutionsEnvironment) common.Executor {
+	return func(ctx context.Context) error {
+		sctx, cancel := context.WithTimeout(ctx, time.Minute*5)
+		defer cancel()
+		health := container.ContainerHealthStarting
+		delay := time.Second
+		for i := 0; ; i++ {
+			health = c.GetHealth(sctx)
+			if health != container.ContainerHealthStarting || i > 30 {
+				break
+			}
+			time.Sleep(delay)
+			delay *= 2
+			if delay > 10*time.Second {
+				delay = 10 * time.Second
+			}
+		}
+		if health == container.ContainerHealthHealthy {
+			return nil
+		}
+		return fmt.Errorf("service container failed to start")
+	}
+}
+
+func (rc *RunContext) waitForServiceContainers() common.Executor {
+	return func(ctx context.Context) error {
+		execs := []common.Executor{}
+		for _, c := range rc.ServiceContainers {
+			execs = append(execs, rc.waitForServiceContainer(c))
+		}
+		return common.NewParallelExecutor(len(execs), execs...)(ctx)
+	}
+}
+
 func (rc *RunContext) stopServiceContainers() common.Executor {
 	return func(ctx context.Context) error {
 		execs := []common.Executor{}

pkg/runner/runner_test.go

@@ -317,6 +317,7 @@ func TestRunEvent(t *testing.T) {
 		{workdir, "services-empty-image", "push", "", platforms, secrets},
 		{workdir, "services-host-network", "push", "", platforms, secrets},
 		{workdir, "services-with-container", "push", "", platforms, secrets},
+		{workdir, "mysql-service-container-with-health-check", "push", "", platforms, secrets},
 
 		// local remote action overrides
 		{workdir, "local-remote-action-overrides", "push", "", platforms, secrets},

pkg/runner/testdata/mysql-service-container-with-health-check/push.yml

@@ -0,0 +1,19 @@
+name: service-container
+on: push
+jobs:
+  service-container-test:
+    runs-on: ubuntu-latest
+    container: mysql:8
+    services:
+      maindb:
+        image: mysql:8
+        env:
+          MYSQL_DATABASE: dbname
+          MYSQL_USER: dbuser
+          MYSQL_PASSWORD: dbpass
+          MYSQL_RANDOM_ROOT_PASSWORD: yes
+        options: --health-cmd="mysqladmin ping" --health-interval=10s --health-timeout=5s --health-retries=3
+    steps:
+    - run: mysql -u dbuser -D dbname -pdbpass -h maindb -e "create table T(id INT NOT NULL AUTO_INCREMENT, val VARCHAR(255), PRIMARY KEY (id))"
+    - run: mysql -u dbuser -D dbname -pdbpass -h maindb -e "insert into T(val) values ('test'),('h')"
+    - run: mysql -u dbuser -D dbname -pdbpass -h maindb -e "select * from T"