PMM-14665 Skip nodes internal to a PMM deployment when adding a service

api/management/v1/azure.proto

@@ -130,6 +130,8 @@ message AddAzureDatabaseRequest {
   google.protobuf.Duration connection_timeout = 26 [(validate.rules).duration = {
     gte: {seconds: 0}
   }];
+  // The pmm-agent identifier which should run agents. Defaults to the PMM Server's own pmm-agent.
+  string pmm_agent_id = 27;
 }
 
 message AddAzureDatabaseResponse {}

api/management/v1/json/v1.json

@@ -803,6 +803,11 @@
                         "description": "True if this node is a PMM Server node (HA mode).",
                         "type": "boolean",
                         "x-order": 18
+                      },
+                      "is_pmm_internal_node": {
+                        "description": "True if this node belongs to the internal infrastructure of a PMM deployment\n(e.g. the HA persistence layer) and must not host user monitoring workloads.",
+                        "type": "boolean",
+                        "x-order": 19
                       }
                     }
                   },
@@ -1360,6 +1365,11 @@
                       "description": "True if this node is a PMM Server node (HA mode).",
                       "type": "boolean",
                       "x-order": 18
+                    },
+                    "is_pmm_internal_node": {
+                      "description": "True if this node belongs to the internal infrastructure of a PMM deployment\n(e.g. the HA persistence layer) and must not host user monitoring workloads.",
+                      "type": "boolean",
+                      "x-order": 19
                     }
                   },
                   "x-order": 0
@@ -7131,6 +7141,11 @@
                   "description": "Connection timeout for exporter (if set).",
                   "type": "string",
                   "x-order": 25
+                },
+                "pmm_agent_id": {
+                  "description": "The pmm-agent identifier which should run agents. Defaults to the PMM Server's own pmm-agent.",
+                  "type": "string",
+                  "x-order": 26
                 }
               }
             }

api/management/v1/node.proto

@@ -165,6 +165,9 @@ message UniversalNode {
   string instance_id = 18;
   // True if this node is a PMM Server node (HA mode).
   bool is_pmm_server_node = 19;
+  // True if this node belongs to the internal infrastructure of a PMM deployment
+  // (e.g. the HA persistence layer) and must not host user monitoring workloads.
+  bool is_pmm_internal_node = 20;
 }
 
 message ListNodesRequest {

api/swagger/swagger-dev.json

@@ -22326,6 +22326,11 @@
                         "description": "True if this node is a PMM Server node (HA mode).",
                         "type": "boolean",
                         "x-order": 18
+                      },
+                      "is_pmm_internal_node": {
+                        "description": "True if this node belongs to the internal infrastructure of a PMM deployment\n(e.g. the HA persistence layer) and must not host user monitoring workloads.",
+                        "type": "boolean",
+                        "x-order": 19
                       }
                     }
                   },
@@ -22883,6 +22888,11 @@
                       "description": "True if this node is a PMM Server node (HA mode).",
                       "type": "boolean",
                       "x-order": 18
+                    },
+                    "is_pmm_internal_node": {
+                      "description": "True if this node belongs to the internal infrastructure of a PMM deployment\n(e.g. the HA persistence layer) and must not host user monitoring workloads.",
+                      "type": "boolean",
+                      "x-order": 19
                     }
                   },
                   "x-order": 0
@@ -28654,6 +28664,11 @@
                   "description": "Connection timeout for exporter (if set).",
                   "type": "string",
                   "x-order": 25
+                },
+                "pmm_agent_id": {
+                  "description": "The pmm-agent identifier which should run agents. Defaults to the PMM Server's own pmm-agent.",
+                  "type": "string",
+                  "x-order": 26
                 }
               }
             }

api/swagger/swagger.json

@@ -21353,6 +21353,11 @@
                         "description": "True if this node is a PMM Server node (HA mode).",
                         "type": "boolean",
                         "x-order": 18
+                      },
+                      "is_pmm_internal_node": {
+                        "description": "True if this node belongs to the internal infrastructure of a PMM deployment\n(e.g. the HA persistence layer) and must not host user monitoring workloads.",
+                        "type": "boolean",
+                        "x-order": 19
                       }
                     }
                   },
@@ -21910,6 +21915,11 @@
                       "description": "True if this node is a PMM Server node (HA mode).",
                       "type": "boolean",
                       "x-order": 18
+                    },
+                    "is_pmm_internal_node": {
+                      "description": "True if this node belongs to the internal infrastructure of a PMM deployment\n(e.g. the HA persistence layer) and must not host user monitoring workloads.",
+                      "type": "boolean",
+                      "x-order": 19
                     }
                   },
                   "x-order": 0
@@ -27681,6 +27691,11 @@
                   "description": "Connection timeout for exporter (if set).",
                   "type": "string",
                   "x-order": 25
+                },
+                "pmm_agent_id": {
+                  "description": "The pmm-agent identifier which should run agents. Defaults to the PMM Server's own pmm-agent.",
+                  "type": "string",
+                  "x-order": 26
                 }
               }
             }

managed/cmd/pmm-managed/main.go

@@ -234,6 +234,20 @@ type gRPCServerDeps struct {
 	versionCache              *versioncache.Service
 	vmdb                      *victoriametrics.Service
 	vmalert                   *vmalert.Service
+	internalNodePrefixes      []string
+}
+
+// parseNodeNamePrefixes splits a comma-separated list of Node name prefixes.
+func parseNodeNamePrefixes(value string) []string {
+	var prefixes []string
+	for p := range strings.SplitSeq(value, ",") {
+		p = strings.TrimSpace(p)
+		if p != "" {
+			prefixes = append(prefixes, p)
+		}
+	}
+
+	return prefixes
 }
 
 // runGRPCServer runs gRPC server until context is canceled, then gracefully stops it.
@@ -304,6 +318,7 @@ func runGRPCServer(ctx context.Context, deps *gRPCServerDeps) {
 		deps.db, deps.agentsRegistry, deps.agentsStateUpdater,
 		deps.connectionCheck, deps.serviceInfoBroker, deps.vmdb,
 		deps.versionCache, deps.grafanaClient, v1.NewAPI(*deps.vmClient),
+		deps.internalNodePrefixes,
 	)
 
 	managementv1.RegisterManagementServiceServer(gRPCServer, managementSvc)
@@ -743,6 +758,11 @@ func main() { //nolint:gocognit,maintidx,cyclop
 		Default("9762").
 		Int()
 
+	internalNodePrefixesF := kingpin.Flag("internal-node-name-prefixes",
+		"Comma-separated list of Node name prefixes reserved for the internal infrastructure of this PMM deployment").
+		Envar("PMM_INTERNAL_NODE_NAME_PREFIXES").
+		String()
+
 	supervisordConfigDirF := kingpin.Flag("supervisord-config-dir", "Supervisord configuration directory").Required().String()
 
 	logLevelF := kingpin.Flag("log-level", "Set logging level").Envar("PMM_LOG_LEVEL").Default("info").Enum("trace", "debug", "info", "warn", "error", "fatal")
@@ -1196,6 +1216,7 @@ func main() { //nolint:gocognit,maintidx,cyclop
 				grafanaClient:             grafanaClient,
 				handler:                   agentsHandler,
 				ha:                        haService,
+				internalNodePrefixes:      parseNodeNamePrefixes(*internalNodePrefixesF),
 				jobsService:               jobsService,
 				minioClient:               minioClient,
 				pbmPITRService:            pbmPITRService,

managed/cmd/pmm-managed/main_test.go

@@ -206,3 +206,17 @@ func formatPkgName(t *testing.T, name string) string {
 
 	return name
 }
+
+func TestParseNodeNamePrefixes(t *testing.T) {
+	for _, tc := range []struct {
+		value    string
+		expected []string
+	}{
+		{value: "", expected: nil},
+		{value: ",,", expected: nil},
+		{value: "pmm-pmm-ha-pg-db-", expected: []string{"pmm-pmm-ha-pg-db-"}},
+		{value: " pmm-pmm-ha-pg-db- , pmm-pmm-ha-ch- ", expected: []string{"pmm-pmm-ha-pg-db-", "pmm-pmm-ha-ch-"}},
+	} {
+		assert.Equal(t, tc.expected, parseNodeNamePrefixes(tc.value), tc.value)
+	}
+}

managed/services/management/add_service_exporter_timeout_test.go

@@ -75,7 +75,7 @@ func TestAddServiceExporterTimeout(t *testing.T) {
 		vmClient.AssertExpectations(t)
 	})
 
-	s := NewManagementService(db, ar, state, cc, sib, vmdb, vc, grafanaClient, vmClient)
+	s := NewManagementService(db, ar, state, cc, sib, vmdb, vc, grafanaClient, vmClient, nil)
 	want := durationpb.New(17 * time.Second)
 
 	t.Run("MySQL", func(t *testing.T) {

managed/services/management/agent_test.go

@@ -97,7 +97,7 @@ func setup(t *testing.T) (context.Context, *ManagementService, func(t *testing.T
 		vmClient.AssertExpectations(t)
 	}
 
-	s := NewManagementService(db, ar, state, cc, sib, vmdb, vc, grafanaClient, vmClient)
+	s := NewManagementService(db, ar, state, cc, sib, vmdb, vc, grafanaClient, vmClient, nil)
 
 	return ctx, s, teardown
 }

managed/services/management/annotation_test.go

@@ -66,7 +66,7 @@ func TestAnnotations(t *testing.T) {
 		vmClient := &mockVictoriaMetricsClient{}
 		vmClient.Test(t)
 
-		s := NewManagementService(db, ar, state, cc, sib, vmdb, vc, grafanaClient, vmClient)
+		s := NewManagementService(db, ar, state, cc, sib, vmdb, vc, grafanaClient, vmClient, nil)
 
 		teardown := func(t *testing.T) {
 			t.Helper()