feat(nvca): wire BYOO metric subset config

deploy/helm/nvca-operator/nvca-operator/values.yaml

@@ -100,6 +100,21 @@ agentConfig:
   # mergeConfig: |
   #   agent:
   #     logLevel: debug
+  #
+  # BYOO metric subset example:
+  #
+  # mergeConfig: |
+  #   agent:
+  #     byooMetricSubset:
+  #       enabled: true
+  #       filterConfig: |
+  #         error_mode: ignore
+  #         metric_conditions:
+  #           - 'metric.name != "BpsInstrument"'
+  #     byooWorkloadMetrics:
+  #       dropLabels:
+  #         - metric_subset_enabled
+  #         - custom_label
 ## @section resources Resource requests and limits for the nvca-operator container
 ## @param resources.limits.cpu CPU limit for the nvca-operator container
 ## @param resources.limits.memory Memory limit for the nvca-operator container

docs/ngc-managed/cluster-management/configuration.md

@@ -733,6 +733,24 @@ pod and merged with the generated config at runtime.
        logLevel: debug
 ```
 
+BYOO metric subset example:
+
+```yaml
+ agentConfig:
+   mergeConfig: |
+     agent:
+       byooMetricSubset:
+         enabled: true
+         filterConfig: |
+           error_mode: ignore
+           metric_conditions:
+             - 'metric.name != "BpsInstrument"'
+       byooWorkloadMetrics:
+         dropLabels:
+           - metric_subset_enabled
+           - custom_label
+```
+
 **Apply via Helm:**
 
 ```bash

docs/user/cluster-management/configuration.md

@@ -854,6 +854,24 @@ isolated test clusters that run the LLM request router tunnel without TLS. For
 the full LLM addon setup, see
 [LLM Function Enablement](../llm-function-enablement.md).
 
+BYOO metric subset example:
+
+```yaml
+agentConfig:
+  mergeConfig: |
+    agent:
+      byooMetricSubset:
+        enabled: true
+        filterConfig: |
+          error_mode: ignore
+          metric_conditions:
+            - 'metric.name != "BpsInstrument"'
+      byooWorkloadMetrics:
+        dropLabels:
+          - metric_subset_enabled
+          - custom_label
+```
+
 Apply via Helm:
 
 ```bash

src/compute-plane-services/byoo-otel-collector/README.md

@@ -104,7 +104,7 @@ Exposed Ports:
 - 14358: OTLP HTTP receiver
 - 13133: `/health?verbose` endpoint to get detailed health status of collector (healthcheck v2 extension)
 - 19090: `/metrics` endpoint for the byoo-otel-collector metrics
-- 19091: `/metrics` endpoint for optional SRE user metrics
+- 19091: `/metrics` endpoint for optional metric subset user metrics
 
 ### nvcf-otel-collector Image
 
@@ -129,9 +129,9 @@ Chunking is disabled by default. Configure it with:
 - `BYOO_LOG_CHUNK_MAX_BODY_BYTES`: maximum log body size in bytes before chunking. `0` disables the processor. Enabled values must be at least `4` bytes so chunks can preserve UTF-8 rune boundaries. Use `983040` bytes for normal BYOO deployments to leave room for log attributes and exporter envelope overhead under a `1000000` byte backend entry limit.
 - `BYOO_LOG_CHUNK_DRY_RUN`: records oversized-log metrics and warnings without mutating log payloads. Dry-run metric datapoints use `mode=dry_run`.
 - `BYOO_LOG_EXPORTER_BATCH_MAX_SIZE_BYTES`: serialized log export request batch size used for exporterhelper byte splitting. `0` or unset uses the default `1000000` bytes.
-- `BYOO_SRE_METRICS_ENABLED`: enables an additional OTLP-only metrics pipeline that exposes filtered user metrics through a Prometheus exporter on port `19091`. Disabled by default.
-- `BYOO_SRE_METRICS_FILTER_CONFIG`: optional YAML filter processor config for the SRE metrics pipeline. If unset, the default drops every metric except `BpsInstrument`, `FpsInstrument`, `RtdInstrument`, and `StageOpenDuration`, and drops datapoints/resources explicitly labeled `sre_metrics_enabled=false` or `sre_enabled_metrics=false`.
-- `BYOO_CUSTOMER_METRICS_DROP_LABELS`: comma-separated resource attribute names removed from the customer metrics pipeline when SRE metrics are enabled. If unset, defaults to `sre_metrics_enabled,sre_enabled_metrics`.
+- `BYOO_METRIC_SUBSET_ENABLED`: enables an additional OTLP-only metrics pipeline that exposes filtered user metrics through a Prometheus exporter on port `19091`. Disabled by default.
+- `BYOO_METRIC_SUBSET_FILTER_CONFIG`: optional YAML filter processor config for the metric subset pipeline. If unset, the default drops every metric except `BpsInstrument`, `FpsInstrument`, `RtdInstrument`, and `StageOpenDuration`, and drops datapoints/resources explicitly labeled `metric_subset_enabled=false`.
+- `BYOO_WORKLOAD_METRICS_DROP_LABELS`: comma-separated resource attribute names removed from the generated workload `metrics` pipeline. If unset, defaults to `metric_subset_enabled` only when the metric subset pipeline is enabled.
 
 When chunking is enabled, each emitted chunk preserves the original log metadata and adds these attributes so chunks can be grouped in the backend:
 

src/compute-plane-services/byoo-otel-collector/examples/otelconfigs/k8s/config_function_container_sre_metrics.yaml → src/compute-plane-services/byoo-otel-collector/examples/otelconfigs/k8s/config_function_container_metric_subset.yaml

@@ -113,11 +113,11 @@ exporters:
     resource_to_telemetry_conversion:
       enabled: true
     send_timestamps: true
-  prometheusremotewrite/PROMETHEUS-customer-metrics-metrics:
-    endpoint: https://customer-metrics.example.invalid/api/v1/write
+  prometheusremotewrite/PROMETHEUS-workload-metrics-metrics:
+    endpoint: https://workload-metrics.example.invalid/api/v1/write
     tls:
-      cert_file: /etc/byoo-otel-collector/secrets/customer-metrics-clientCert
-      key_file: /etc/byoo-otel-collector/secrets/customer-metrics-clientKey
+      cert_file: /etc/byoo-otel-collector/secrets/workload-metrics-clientCert
+      key_file: /etc/byoo-otel-collector/secrets/workload-metrics-clientKey
 processors:
   attributes/add-metadata:
     actions:
@@ -149,10 +149,16 @@ processors:
     send_batch_max_size: 8192
     send_batch_size: 4096
     timeout: 400ms
-  batch/sre_metrics:
+  batch/metric_subset:
     send_batch_max_size: 8192
     send_batch_size: 4096
     timeout: 400ms
+  filter/metric_subset:
+    error_mode: ignore
+    metric_conditions:
+      - metric.name != "BpsInstrument" and metric.name != "FpsInstrument" and metric.name != "RtdInstrument" and metric.name != "StageOpenDuration"
+      - resource.attributes["metric_subset_enabled"] == "false"
+      - datapoint.attributes["metric_subset_enabled"] == "false"
   filter/metrics:
     metrics:
       exclude:
@@ -162,14 +168,6 @@ processors:
           - scrape_samples_post_metric_relabeling
           - scrape_samples_scraped
           - scrape_series_added
-  filter/sre_metrics:
-    error_mode: ignore
-    metric_conditions:
-      - metric.name != "BpsInstrument" and metric.name != "FpsInstrument" and metric.name != "RtdInstrument" and metric.name != "StageOpenDuration"
-      - resource.attributes["sre_metrics_enabled"] == "false"
-      - resource.attributes["sre_enabled_metrics"] == "false"
-      - datapoint.attributes["sre_metrics_enabled"] == "false"
-      - datapoint.attributes["sre_enabled_metrics"] == "false"
   memory_limiter:
     check_interval: 1s
     limit_percentage: 80
@@ -208,12 +206,10 @@ processors:
     attributes:
       - action: delete
         key: service.instance.id
-  resource/customer_metrics:
+  resource/workload_metrics_drop_labels:
     attributes:
       - action: delete
-        key: sre_metrics_enabled
-      - action: delete
-        key: sre_enabled_metrics
+        key: metric_subset_enabled
 extensions:
   cgroup_runtime:
     gomaxprocs:
@@ -274,25 +270,25 @@ service:
         - otlp
         - prometheus
       exporters:
-        - prometheusremotewrite/PROMETHEUS-customer-metrics-metrics
+        - prometheusremotewrite/PROMETHEUS-workload-metrics-metrics
       processors:
         - memory_limiter
         - filter/metrics
         - resource
-        - resource/customer_metrics
+        - resource/workload_metrics_drop_labels
         - metrics_transform
         - batch
-    metrics/sre:
+    metrics/metric_subset:
       receivers:
         - otlp
       exporters:
         - prometheus/user-metrics
       processors:
         - memory_limiter
-        - filter/sre_metrics
+        - filter/metric_subset
         - resource
         - metrics_transform
-        - batch/sre_metrics
+        - batch/metric_subset
   extensions:
     - healthcheckv2
     - cgroup_runtime

src/compute-plane-services/byoo-otel-collector/internal/otelconfig/embed_config.go

@@ -47,7 +47,8 @@ type TemplateConfig struct {
 	InstanceID        string
 	ZoneName          string
 	LogChunking       LogChunkingConfig
-	SREMetrics        SREMetricsConfig
+	MetricSubset      MetricSubsetConfig
+	WorkloadMetrics   WorkloadMetricsConfig
 	// LogExporterBatchMaxSizeBytes configures exporterhelper byte batching for logs.
 	// Zero uses the default selected for BYOO.
 	LogExporterBatchMaxSizeBytes int
@@ -58,10 +59,13 @@ type LogChunkingConfig struct {
 	DryRun       bool
 }
 
-type SREMetricsConfig struct {
-	Enabled                   bool
-	FilterConfig              map[string]interface{}
-	CustomerMetricsDropLabels []string
+type WorkloadMetricsConfig struct {
+	DropLabels []string
+}
+
+type MetricSubsetConfig struct {
+	Enabled      bool
+	FilterConfig map[string]interface{}
 }
 
 func ExecuteTemplate(w io.Writer, tcfg TemplateConfig) error {

src/compute-plane-services/byoo-otel-collector/internal/otelconfig/otelconfig.go

@@ -42,9 +42,9 @@ type envConfig struct {
 	ByooLogChunkMaxBodyBytes         int    `split_words:"true"`
 	ByooLogChunkDryRun               bool   `split_words:"true"`
 	ByooLogExporterBatchMaxSizeBytes int    `split_words:"true"`
-	ByooSreMetricsEnabled            bool   `split_words:"true"`
-	ByooSreMetricsFilterConfig       string `split_words:"true"`
-	ByooCustomerMetricsDropLabels    string `split_words:"true"`
+	ByooMetricSubsetEnabled          bool   `split_words:"true"`
+	ByooMetricSubsetFilterConfig     string `split_words:"true"`
+	ByooWorkloadMetricsDropLabels    string `split_words:"true"`
 }
 
 func processEnvConfig(env *envConfig) error {
@@ -79,15 +79,15 @@ func getTemplateConfig() (TemplateConfig, error) {
 		return TemplateConfig{}, fmt.Errorf("BYOO_LOG_EXPORTER_BATCH_MAX_SIZE_BYTES: %w", err)
 	}
 	tcgf.LogExporterBatchMaxSizeBytes = logExporterBatchMaxSizeBytes
-	sreMetricsFilterConfig, err := resolvedSREMetricsFilterConfig(env.ByooSreMetricsFilterConfig)
+	metricSubsetFilterConfig, err := resolvedMetricSubsetFilterConfig(env.ByooMetricSubsetFilterConfig)
 	if err != nil {
-		return TemplateConfig{}, fmt.Errorf("BYOO_SRE_METRICS_FILTER_CONFIG: %w", err)
+		return TemplateConfig{}, fmt.Errorf("BYOO_METRIC_SUBSET_FILTER_CONFIG: %w", err)
 	}
-	tcgf.SREMetrics = SREMetricsConfig{
-		Enabled:                   env.ByooSreMetricsEnabled,
-		FilterConfig:              sreMetricsFilterConfig,
-		CustomerMetricsDropLabels: resolvedCustomerMetricsDropLabels(env.ByooCustomerMetricsDropLabels),
+	tcgf.MetricSubset = MetricSubsetConfig{
+		Enabled:      env.ByooMetricSubsetEnabled,
+		FilterConfig: metricSubsetFilterConfig,
 	}
+	tcgf.WorkloadMetrics.DropLabels = resolvedWorkloadMetricsDropLabels(env.ByooWorkloadMetricsDropLabels, env.ByooMetricSubsetEnabled)
 
 	functionID := env.NvcfFunctionID
 	functionVersionID := env.NvcfFunctionVersionID

src/compute-plane-services/byoo-otel-collector/internal/otelconfig/otelconfig_test.go

@@ -117,40 +117,40 @@ func TestGetTemplateConfig(t *testing.T) {
 			},
 		},
 		{
-			name: "custom SRE metrics config",
+			name: "custom metric subset config",
 			env: map[string]string{
 				"NVCF_BACKEND_TYPE":                 "gfn",
 				"NVCF_INSTANCE_ID":                  "test-instance",
 				"NVCF_NAMESPACE":                    "test-ns",
 				"NVCF_WORKLOAD_TYPE":                "function",
 				"NVCT_TASK_ID":                      "task-123",
 				"NVCF_ZONE_NAME":                    "zone-1",
-				"BYOO_SRE_METRICS_ENABLED":          "true",
-				"BYOO_SRE_METRICS_FILTER_CONFIG":    "error_mode: ignore\nmetric_conditions:\n  - 'metric.name == \"drop\"'\n",
-				"BYOO_CUSTOMER_METRICS_DROP_LABELS": "sre_metrics_enabled, custom_label, sre_metrics_enabled",
+				"BYOO_METRIC_SUBSET_ENABLED":        "true",
+				"BYOO_METRIC_SUBSET_FILTER_CONFIG":  "error_mode: ignore\nmetric_conditions:\n  - 'metric.name == \"drop\"'\n",
+				"BYOO_WORKLOAD_METRICS_DROP_LABELS": "metric_subset_enabled, custom_label, metric_subset_enabled",
 			},
 			expectErr: false,
 			expect: func(t *testing.T, cfg TemplateConfig) {
-				assert.True(t, cfg.SREMetrics.Enabled)
+				assert.True(t, cfg.MetricSubset.Enabled)
 				assert.Equal(t, map[string]interface{}{
 					"error_mode": "ignore",
 					"metric_conditions": []interface{}{
 						`metric.name == "drop"`,
 					},
-				}, cfg.SREMetrics.FilterConfig)
-				assert.Equal(t, []string{"sre_metrics_enabled", "custom_label"}, cfg.SREMetrics.CustomerMetricsDropLabels)
+				}, cfg.MetricSubset.FilterConfig)
+				assert.Equal(t, []string{"metric_subset_enabled", "custom_label"}, cfg.WorkloadMetrics.DropLabels)
 			},
 		},
 		{
-			name: "invalid SRE metrics filter config",
+			name: "invalid metric subset filter config",
 			env: map[string]string{
-				"NVCF_BACKEND_TYPE":              "gfn",
-				"NVCF_INSTANCE_ID":               "test-instance",
-				"NVCF_NAMESPACE":                 "test-ns",
-				"NVCF_WORKLOAD_TYPE":             "function",
-				"NVCT_TASK_ID":                   "task-123",
-				"NVCF_ZONE_NAME":                 "zone-1",
-				"BYOO_SRE_METRICS_FILTER_CONFIG": "processors: []",
+				"NVCF_BACKEND_TYPE":                "gfn",
+				"NVCF_INSTANCE_ID":                 "test-instance",
+				"NVCF_NAMESPACE":                   "test-ns",
+				"NVCF_WORKLOAD_TYPE":               "function",
+				"NVCT_TASK_ID":                     "task-123",
+				"NVCF_ZONE_NAME":                   "zone-1",
+				"BYOO_METRIC_SUBSET_FILTER_CONFIG": "processors: []",
 			},
 			expectErr: true,
 		},
@@ -240,9 +240,9 @@ func TestGetTemplateConfig(t *testing.T) {
 				"BYOO_LOG_CHUNK_MAX_BODY_BYTES",
 				"BYOO_LOG_CHUNK_DRY_RUN",
 				"BYOO_LOG_EXPORTER_BATCH_MAX_SIZE_BYTES",
-				"BYOO_SRE_METRICS_ENABLED",
-				"BYOO_SRE_METRICS_FILTER_CONFIG",
-				"BYOO_CUSTOMER_METRICS_DROP_LABELS",
+				"BYOO_METRIC_SUBSET_ENABLED",
+				"BYOO_METRIC_SUBSET_FILTER_CONFIG",
+				"BYOO_WORKLOAD_METRICS_DROP_LABELS",
 			}
 			backup := map[string]*string{}
 			for _, k := range envKeys {