feat: preare batches before routing

Signed-off-by: Dominik Rosiek <[email protected]>
SumoLogic · Jul 4, 2023 · d373d1c · d373d1c
1 parent 6e2b05c
commit d373d1c
Show file tree

Hide file tree

Showing 5 changed files with 77 additions and 14 deletions.
diff --git a/deploy/helm/sumologic/conf/metrics/otelcol/pipeline.yaml b/deploy/helm/sumologic/conf/metrics/otelcol/pipeline.yaml
@@ -22,6 +22,9 @@ processors:
   - resource/remove_k8s_pod_pod_name
   - resource/delete_source_metadata
   - sumologic_schema
+  - transform/set_name
+  - groupbyattrs/group_by_name
+  - transform/remove_name
   - batch
   - transform/prepare_routing
   - routing

diff --git a/deploy/helm/sumologic/conf/metrics/otelcol/processors.yaml b/deploy/helm/sumologic/conf/metrics/otelcol/processors.yaml
@@ -16,6 +16,11 @@ groupbyattrs:
     - pod
     - service
 
+groupbyattrs/group_by_name:
+  keys:
+    - __name__
+    - job
+
 ## The Kubernetes sprocessor automatically tags logs, metrics and traces with Kubernetes metadata like pod name, namespace name etc.
 k8s_tagger:
   extract:
@@ -110,20 +115,6 @@ resource/remove_k8s_pod_pod_name:
     - action: delete
       key: k8s.pod.pod_name
 
-transform/prepare_routing:
-  error_mode: ignore
-  metric_statements:
-    - context: metric
-      statements:
-        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.apiserver") where resource.attributes["job"] == "apiserver"
-        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.control-plane.coredns") where resource.attributes["job"] == "coredns"
-        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.control-plane.kube-etcd") where resource.attributes["job"] == "kube-etcd"
-        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.controller-manager") where resource.attributes["job"] == "kubelet" and IsMatch(name, "^cloudprovider_.*")
-        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.kubelet") where resource.attributes["job"] == "kubelet" and not IsMatch(name, "^cloudprovider_.*")
-        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.node") where resource.attributes["job"] == "node-exporter"
-        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.scheduler") where resource.attributes["job"] == "kube-scheduler"
-        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.state") where resource.attributes["job"] == "kube-state-metrics"
-
 ## NOTE: below listed rules could be simplified if routingprocessor
 ## supports regex matching. At this point we could group route entries
 ## going to the same set of exporters.
@@ -176,6 +167,34 @@ source:
 sumologic_schema:
   add_cloud_namespace: false
 
+transform/prepare_routing:
+  error_mode: ignore
+  metric_statements:
+    - context: metric
+      statements:
+        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.apiserver") where resource.attributes["job"] == "apiserver"
+        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.control-plane.coredns") where resource.attributes["job"] == "coredns"
+        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.control-plane.kube-etcd") where resource.attributes["job"] == "kube-etcd"
+        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.controller-manager") where resource.attributes["job"] == "kubelet" and IsMatch(name, "^cloudprovider_.*")
+        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.kubelet") where resource.attributes["job"] == "kubelet" and not IsMatch(name, "^cloudprovider_.*")
+        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.node") where resource.attributes["job"] == "node-exporter"
+        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.scheduler") where resource.attributes["job"] == "kube-scheduler"
+        - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.state") where resource.attributes["job"] == "kube-state-metrics"
+
+transform/remove_name:
+  error_mode: ignore
+  metric_statements:
+    - context: resource
+      statements:
+        - delete_key(attributes, "__name__")
+
+transform/set_name:
+  error_mode: ignore
+  metric_statements:
+    - context: datapoint
+      statements:
+        - set(attributes["__name__"], metric.name) where IsMatch(metric.name, "^cloudprovider_.*")
+
 {{- if .Values.sumologic.metrics.otelcol.extraProcessors }}
 {{- range $processor := .Values.sumologic.metrics.otelcol.extraProcessors }}
 {{ toYaml $processor }}

diff --git a/tests/helm/metrics_test.go b/tests/helm/metrics_test.go
@@ -131,6 +131,9 @@ func TestMetadataMetricsOtelConfigExtraProcessors(t *testing.T) {
 		"resource/remove_k8s_pod_pod_name",
 		"resource/delete_source_metadata",
 		"sumologic_schema",
+		"transform/set_name",
+		"groupbyattrs/group_by_name",
+		"transform/remove_name",
 		"batch",
 		"transform/prepare_routing",
 		"routing",

diff --git a/tests/helm/testdata/goldenfile/metadata_metrics_otc/additional_endpoints.output.yaml b/tests/helm/testdata/goldenfile/metadata_metrics_otc/additional_endpoints.output.yaml
@@ -113,6 +113,10 @@ data:
         - namespace
         - pod
         - service
+      groupbyattrs/group_by_name:
+        keys:
+        - __name__
+        - job
       k8s_tagger:
         extract:
           delimiter: _
@@ -241,6 +245,18 @@ data:
             where resource.attributes["job"] == "kube-scheduler"
           - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.state")
             where resource.attributes["job"] == "kube-state-metrics"
+      transform/remove_name:
+        error_mode: ignore
+        metric_statements:
+        - context: resource
+          statements:
+          - delete_key(attributes, "__name__")
+      transform/set_name:
+        error_mode: ignore
+        metric_statements:
+        - context: datapoint
+          statements:
+          - set(attributes["__name__"], metric.name) where IsMatch(metric.name, "^cloudprovider_.*")
     receivers:
       otlp:
         protocols:
@@ -321,6 +337,9 @@ data:
           - resource/remove_k8s_pod_pod_name
           - resource/delete_source_metadata
           - sumologic_schema
+          - transform/set_name
+          - groupbyattrs/group_by_name
+          - transform/remove_name
           - batch
           - transform/prepare_routing
           - routing

diff --git a/tests/helm/testdata/goldenfile/metadata_metrics_otc/basic.output.yaml b/tests/helm/testdata/goldenfile/metadata_metrics_otc/basic.output.yaml
@@ -113,6 +113,10 @@ data:
         - namespace
         - pod
         - service
+      groupbyattrs/group_by_name:
+        keys:
+        - __name__
+        - job
       k8s_tagger:
         extract:
           delimiter: _
@@ -241,6 +245,18 @@ data:
             where resource.attributes["job"] == "kube-scheduler"
           - set(resource.attributes["http_listener_v2_path"], "/prometheus.metrics.state")
             where resource.attributes["job"] == "kube-state-metrics"
+      transform/remove_name:
+        error_mode: ignore
+        metric_statements:
+        - context: resource
+          statements:
+          - delete_key(attributes, "__name__")
+      transform/set_name:
+        error_mode: ignore
+        metric_statements:
+        - context: datapoint
+          statements:
+          - set(attributes["__name__"], metric.name) where IsMatch(metric.name, "^cloudprovider_.*")
     receivers:
       otlp:
         protocols:
@@ -319,6 +335,9 @@ data:
           - resource/remove_k8s_pod_pod_name
           - resource/delete_source_metadata
           - sumologic_schema
+          - transform/set_name
+          - groupbyattrs/group_by_name
+          - transform/remove_name
           - batch
           - transform/prepare_routing
           - routing