Stop returning ID outside of the runtime layer, fix k8s ID bug (#1207)

Currently a number of the runtime interface methods expect a container ID to be passed. The format and meaning of this ID is specific to the runtime. Callers of the runtime interface end up having to call the runtime to get an ID and pass it back.

This changes all these methods to require a workload name instead of a container ID. The runtime methods will do a lookup if needed to find the container ID. This hides the runtime-specific IDs from the code outside of the runtime, and maintains the abstraction that workloads are uniquely identified by name.

While implementing this, I noticed a bug where the k8s runtime code gets mixed up between pod IDs and stateful set names (which are the same as the name of the workload). I changed the code to use the workload name to look up stateful sets, and changed the code which operated on pods to search by name and label instead of pod ID.

Author: dmjb

cmd/thv/app/inspector.go

@@ -122,7 +122,7 @@ func inspectorCmdFunc(cmd *cobra.Command, args []string) error {
 	labelsMap := map[string]string{}
 	labels.AddStandardLabels(labelsMap, "inspector", "inspector", string(types.TransportTypeInspector), inspectorUIPort)
 	labelsMap["toolhive-auxiliary"] = "true"
-	_, _, err = rt.DeployWorkload(
+	_, err = rt.DeployWorkload(
 		ctx,
 		processedImage,
 		"inspector",

cmd/thv/app/logs.go

@@ -12,6 +12,7 @@ import (
 	"github.com/spf13/cobra"
 	"github.com/spf13/viper"
 
+	rt "github.com/stacklok/toolhive/pkg/container/runtime"
 	"github.com/stacklok/toolhive/pkg/logger"
 	"github.com/stacklok/toolhive/pkg/workloads"
 )
@@ -71,7 +72,7 @@ func logsCmdFunc(cmd *cobra.Command, args []string) error {
 
 	logs, err := manager.GetLogs(ctx, workloadName, follow)
 	if err != nil {
-		if errors.Is(err, workloads.ErrWorkloadNotFound) {
+		if errors.Is(err, rt.ErrWorkloadNotFound) {
 			logger.Infof("workload %s not found", workloadName)
 			return nil
 		}

cmd/thv/app/stop.go

@@ -7,6 +7,7 @@ import (
 	"github.com/spf13/cobra"
 	"golang.org/x/sync/errgroup"
 
+	rt "github.com/stacklok/toolhive/pkg/container/runtime"
 	"github.com/stacklok/toolhive/pkg/workloads"
 )
 
@@ -97,7 +98,7 @@ func stopCmdFunc(cmd *cobra.Command, args []string) error {
 		group, err = workloadManager.StopWorkloads(ctx, []string{workloadName})
 		if err != nil {
 			// If the workload is not found or not running, treat as a non-fatal error.
-			if errors.Is(err, workloads.ErrWorkloadNotFound) ||
+			if errors.Is(err, rt.ErrWorkloadNotFound) ||
 				errors.Is(err, workloads.ErrWorkloadNotRunning) ||
 				errors.Is(err, workloads.ErrInvalidWorkloadName) {
 				fmt.Printf("workload %s is not running\n", workloadName)

pkg/api/v1/workloads.go

@@ -103,7 +103,7 @@ func (s *WorkloadRoutes) getWorkload(w http.ResponseWriter, r *http.Request) {
 
 	workload, err := s.workloadManager.GetWorkload(ctx, name)
 	if err != nil {
-		if errors.Is(err, workloads.ErrWorkloadNotFound) {
+		if errors.Is(err, runtime.ErrWorkloadNotFound) {
 			http.Error(w, "Workload not found", http.StatusNotFound)
 			return
 		} else if errors.Is(err, workloads.ErrInvalidWorkloadName) {
@@ -429,7 +429,7 @@ func (s *WorkloadRoutes) getLogsForWorkload(w http.ResponseWriter, r *http.Reque
 
 	logs, err := s.workloadManager.GetLogs(ctx, name, false)
 	if err != nil {
-		if errors.Is(err, workloads.ErrWorkloadNotFound) {
+		if errors.Is(err, runtime.ErrWorkloadNotFound) {
 			http.Error(w, "Workload not found", http.StatusNotFound)
 			return
 		}