enhance(k8s): reach Kubernetes services by their id when it is a DNS subdomain (#1320)

On Kubernetes a service is reached by its id made a DNS label, so some service ids might become unreachable. For example, the `devstoreaccount1.azurite.local` service of Gitea's [MSSQL tests](https://github.com/go-gitea/gitea/blob/2fe74d3cdd5184a7e22e32e47772caa78d077489/.github/workflows/pull-db-tests.yml#L229) is only in the pod's hosts as `devstoreaccount1-azurite-local`, while the tests connect to the id as written. Container names must be DNS labels, but `hostAliases` accepts DNS subdomains, so an id that is one is now added there too, and to `no_proxy`, as on Docker where a service is reached by its id.

The warning for a service not reachable by its id now also covers ids that differ only in case, as Kubernetes hostnames are lowercase.

---------

Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1320
Reviewed-by: Lunny Xiao <[email protected]>
Reviewed-by: bircni <[email protected]>
Reviewed-by: silverwind <[email protected]>
Co-authored-by: Zettat123 <[email protected]>

act/container/kubernetes.go

@@ -93,17 +93,29 @@ func (p *KubernetesPod) JobContainer(input *NewContainerInput) ExecutionsEnviron
 	return p.job
 }
 
-var invalidLabelChars = regexp.MustCompile(`[^a-z0-9-]+`)
+var (
+	invalidLabelChars = regexp.MustCompile(`[^a-z0-9-]+`)
+	dnsSubdomain      = regexp.MustCompile(`^[a-z0-9]([-a-z0-9]*[a-z0-9])?(\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*$`)
+)
 
-// KubernetesServiceName returns the container name and hostname of service id, made a DNS label as both must be.
+// kubernetesServiceName returns the container name of service id, made a DNS label as it must be.
-func KubernetesServiceName(id string) string {
+func kubernetesServiceName(id string) string {
 	name := strings.Trim(invalidLabelChars.ReplaceAllString(strings.ToLower(id), "-"), "-")
 	return strings.TrimRight(name[:min(63, len(name))], "-")
 }
 
-// ServiceContainer returns a container of the pod the job reaches at hostname KubernetesServiceName(id).
+// KubernetesServiceHostnames returns the hostnames of service id, its container name and, where hostAliases accept it as a DNS subdomain, id lowercased.
+func KubernetesServiceHostnames(id string) []string {
+	name, host := kubernetesServiceName(id), strings.ToLower(id)
+	if host != name && len(host) <= 253 && dnsSubdomain.MatchString(host) { // https://github.com/kubernetes/apimachinery/blob/v0.35.1/pkg/util/validation/validation.go#L198-L207
+		return []string{name, host}
+	}
+	return []string{name}
+}
+
+// ServiceContainer returns a container of the pod the job reaches at KubernetesServiceHostnames(id).
 func (p *KubernetesPod) ServiceContainer(id string, input *NewContainerInput) ExecutionsEnvironment {
-	service := &kubernetesContainer{pod: p, id: id, name: KubernetesServiceName(id), input: input}
+	service := &kubernetesContainer{pod: p, id: id, name: kubernetesServiceName(id), input: input}
 	p.services = append(p.services, service)
 	return service
 }
@@ -131,7 +143,7 @@ func (p *KubernetesPod) create(ctx context.Context) error {
 		return err
 	}
 	for _, service := range p.services {
-		if !strings.EqualFold(service.id, service.name) {
+		if !slices.ContainsFunc(KubernetesServiceHostnames(service.id), func(hostname string) bool { return strings.EqualFold(hostname, service.id) }) { // glibc and Go match /etc/hosts case-insensitively
 			common.Logger(ctx).Warnf("Service %s is reachable as %s or localhost, as a Kubernetes hostname cannot be %q", service.id, service.name, service.id)
 		}
 	}
@@ -165,7 +177,8 @@ func (p *KubernetesPod) manifest(secret string, secretData map[string]string, me
 		serviceTemplate = templateContainers[index]
 		templateSpec["containers"] = slices.Delete(slices.Clone(templateContainers), index, index+1)
 	}
-	var containers, hostnames, volumes []any
+	var containers, volumes []any
+	var hostnames []string
 	names := map[string]bool{}
 	for _, container := range templateContainers {
 		names[manifestField(container, "name")] = true
@@ -184,7 +197,7 @@ func (p *KubernetesPod) manifest(secret string, secretData map[string]string, me
 			continue
 		}
 		containers = append(containers, mergeManifest(serviceTemplate, spec, ""))
-		hostnames = append(hostnames, c.name)
+		hostnames = append(hostnames, KubernetesServiceHostnames(c.id)...)
 	}
 	for _, name := range []string{"workspace", "act"} {
 		if !slices.ContainsFunc(templateVolumes, func(volume any) bool { return manifestField(volume, "name") == name }) {

act/container/kubernetes_test.go

@@ -129,6 +129,19 @@ func serveLocalExec(t *testing.T, w http.ResponseWriter, r *http.Request) {
 	}
 }
 
+func TestKubernetesServiceHostnames(t *testing.T) {
+	for id, hostnames := range map[string][]string{
+		"db":                             {"db"},
+		"My_DB":                          {"my-db"},
+		"devstoreaccount1.azurite.local": {"devstoreaccount1-azurite-local", "devstoreaccount1.azurite.local"},
+		"DB.Local":                       {"db-local", "db.local"},
+		strings.Repeat("a", 64):          {strings.Repeat("a", 63), strings.Repeat("a", 64)},
+		strings.Repeat("a", 254):         {strings.Repeat("a", 63)},
+	} {
+		assert.Equal(t, hostnames, KubernetesServiceHostnames(id), id)
+	}
+}
+
 func TestKubernetesPodRunsJobThroughAPI(t *testing.T) {
 	if runtime.GOOS == "windows" {
 		t.Skip("uses POSIX shell")

docs/kubernetes.md

@@ -51,7 +51,7 @@ Job pods mount no service account token unless the template sets `automountServi
 
 ## Services
 
-Services share the pod's network, so a step reaches one by its id or `localhost`, and two services cannot listen on the same port. An id is lowercased and has characters other than letters, digits and `-` replaced by `-`, so the service `my_db` is reached as `my-db`. Pass secrets to services through `env`, as `command` is visible in the pod spec. Of `options`, only `--health-*` is supported. They become a startup probe with Docker's defaults, which stops a service that fails its retries, so the job fails as on Docker.
+Services share the pod's network, so a step reaches one by its id or `localhost`, and two services cannot listen on the same port. An id that is not a valid hostname is sanitized, so `my_db` is reached as `my-db`. Pass secrets to services through `env`, as `command` is visible in the pod spec. Of `options`, only `--health-*` is supported. They become a startup probe with Docker's defaults, which stops a service that fails its retries, so the job fails as on Docker.
 
 ## Unsupported
 

internal/app/run/runner.go

@@ -501,9 +501,11 @@ func (r *Runner) run(ctx context.Context, task *runnerv1.Task, reporter *report.
 	_ = job.RawServices.Decode(&services) // a whole-value expression names no service before evaluation
 	serviceNames := slices.Sorted(maps.Keys(services))
 	if strings.HasPrefix(platform, labels.SchemeKubernetes+"://") {
-		for index, name := range serviceNames {
+		var hostnames []string
-			serviceNames[index] = container.KubernetesServiceName(name)
+		for _, name := range serviceNames {
+			hostnames = append(hostnames, container.KubernetesServiceHostnames(name)...)
 		}
+		serviceNames = hostnames
 	}
 	proxyEnv := JobProxyEnv(envs, cacheURL, serviceNames)
 	maps.Copy(envs, proxyEnv)