sanitizer: preserve valid build farm cluster assignments during validation

pkg/sanitizer/determinize.go

@@ -142,6 +142,12 @@ func isCIOperatorLatest(image string) bool {
 
 func determineCluster(jb prowconfig.JobBase, config *dispatcher.Config, pjs map[string]dispatcher.ProwJobData, path string, mostUsedCluster string, blocked sets.Set[string], cm dispatcher.ClusterMap) (string, error) {
 	if pjs == nil {
+		// If the job already has a cluster assigned, check if it's a valid build farm cluster
+		// that's not blocked. If so, preserve the existing assignment to avoid
+		// inconsistencies between what the dispatcher assigned and what this function would calculate.
+		if jb.Cluster != "" && !blocked.Has(jb.Cluster) && config.IsInBuildFarm(api.Cluster(jb.Cluster)) != "" {
+			return jb.Cluster, nil
+		}
 		c, canBeRelocated, err := config.DetermineClusterForJob(jb, path, cm)
 		if err != nil {
 			return "", err

pkg/sanitizer/determinize_test.go

@@ -6,6 +6,7 @@ import (
 	"k8s.io/apimachinery/pkg/util/sets"
 	"sigs.k8s.io/prow/pkg/config"
 
+	"github.com/openshift/ci-tools/pkg/api"
 	"github.com/openshift/ci-tools/pkg/dispatcher"
 )
 
@@ -48,6 +49,95 @@ func TestDefaultJobConfig(t *testing.T) {
 	}
 }
 
+func TestDetermineClusterPreservesValidBuildFarmCluster(t *testing.T) {
+	testCases := []struct {
+		name            string
+		jobCluster      string
+		blocked         sets.Set[string]
+		buildFarm       map[api.Cloud]map[api.Cluster]*dispatcher.BuildFarmConfig
+		groups          dispatcher.JobGroups
+		expectedCluster string
+	}{
+		{
+			name:       "preserves existing valid build farm cluster",
+			jobCluster: "build01",
+			blocked:    sets.New[string](),
+			buildFarm: map[api.Cloud]map[api.Cluster]*dispatcher.BuildFarmConfig{
+				api.CloudAWS: {
+					"build01": {Filenames: sets.New[string]()},
+					"build02": {Filenames: sets.New[string]()},
+				},
+			},
+			expectedCluster: "build01",
+		},
+		{
+			name:       "uses algorithm when cluster is blocked",
+			jobCluster: "build01",
+			blocked:    sets.New[string]("build01"),
+			buildFarm: map[api.Cloud]map[api.Cluster]*dispatcher.BuildFarmConfig{
+				api.CloudAWS: {
+					"build01": {Filenames: sets.New[string]()},
+					"build02": {Filenames: sets.New[string]()},
+				},
+			},
+			groups: dispatcher.JobGroups{
+				"special-cluster": {Jobs: []string{"test-job"}},
+			},
+			expectedCluster: "special-cluster", // algorithm matches job name in Groups
+		},
+		{
+			name:       "uses algorithm when cluster is not in build farm",
+			jobCluster: "some-other-cluster",
+			blocked:    sets.New[string](),
+			buildFarm: map[api.Cloud]map[api.Cluster]*dispatcher.BuildFarmConfig{
+				api.CloudAWS: {
+					"build01": {Filenames: sets.New[string]()},
+					"build02": {Filenames: sets.New[string]()},
+				},
+			},
+			groups: dispatcher.JobGroups{
+				"special-cluster": {Jobs: []string{"test-job"}},
+			},
+			expectedCluster: "special-cluster", // algorithm matches job name in Groups
+		},
+		{
+			name:       "uses algorithm when cluster is empty",
+			jobCluster: "",
+			blocked:    sets.New[string](),
+			buildFarm:  map[api.Cloud]map[api.Cluster]*dispatcher.BuildFarmConfig{},
+			groups: dispatcher.JobGroups{
+				"special-cluster": {Jobs: []string{"test-job"}},
+			},
+			expectedCluster: "special-cluster", // algorithm matches job name in Groups
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			jc := &config.JobConfig{
+				PresubmitsStatic: map[string][]config.Presubmit{
+					"org/repo": {{JobBase: config.JobBase{Agent: "kubernetes", Cluster: tc.jobCluster, Name: "test-job"}}},
+				},
+			}
+
+			cfg := &dispatcher.Config{
+				Default:   "api.ci",
+				BuildFarm: tc.buildFarm,
+				Groups:    tc.groups,
+			}
+
+			if err := defaultJobConfig(jc, "org/repo/test.yaml", cfg, nil, tc.blocked, dispatcher.ClusterMap{}); err != nil {
+				t.Fatalf("failed to default job config: %v", err)
+			}
+
+			got := jc.PresubmitsStatic["org/repo"][0].Cluster
+			if got != tc.expectedCluster {
+				t.Errorf("expected cluster %q, got %q", tc.expectedCluster, got)
+			}
+		})
+	}
+}
+
 func TestIsCIOperatorLatest(t *testing.T) {
 	testCases := []struct {
 		name     string