New feature 'Automatic Prometheus Collection' and upgrade agent image to version 1.8

charts/mgmt-agent/README.md

@@ -25,6 +25,7 @@ A Helm chart for collecting Kubernetes Metrics using OCI Management Agent into O
 | kubernetesCluster.overrideAllowMetricsCluster | string | `nil` | Provide the specific list of comma separated metric names for agent computed metrics to be collected |
 | kubernetesCluster.overrideAllowMetricsKubelet | string | `nil` | Provide the specific list of comma separated metric names for Kubelet (/api/v1/nodes/<node_name>/proxy/metrics) metrics to be collected |
 | kubernetesCluster.overrideAllowMetricsNode | string | `nil` | Provide the specific list of comma separated metric names for Node (/api/v1/nodes/<node_name>/proxy/metrics/resource, /api/v1/nodes/<node_name>/proxy/metrics/cadvisor) metrics to be collected |
+| kubernetesCluster.enableAutomaticPrometheusDetection | bool | `false` | Setting this to true will enable automatic PrometheusEmitter metrics collection from eligible pods |
 | mgmtagent.image.secret | string | `nil` | Image secrets to use for pulling container image (base64 encoded content of ~/.docker/config.json file) |
 | mgmtagent.image.url | string | `nil` | Replace this value with actual docker image URL for Management Agent |
 | mgmtagent.installKey | string | `"resources/input.rsp"` | Copy the downloaded Management Agent Install Key file under root helm directory as resources/input.rsp |

charts/mgmt-agent/resources/auto_prometheus_collection_readme.md

@@ -0,0 +1,77 @@
+# 1. Introduction
+Automatic Prometheus collection allows the Agent to automatically find and identify metrics emitting pods to monitor, eliminating the need to manually create the Prometheus configuration to collect metrics.
+
+# 2. Identification of metric collection configuration
+
+## 2.1 Out of the box
+Management Agent has capabilities to detect metrics emitting pods, without making any custom configuration changes.
+
+### 2.1.1 Pod' ports specification
+Management Agent will look for pod port spec with port name as `metrics` and port protocol as `TCP`.</br>
+Once found, the configuration is build using default path as `/metrics`. The rest of the configuration is set to default as well.
+
+### 2.1.2 Prometheus.io Annotation
+Deployments using industry standard prometheus.io annotations to enable scrapping, will be automatically detected out of the box by agent.</br>
+The pod's ports specification configuration can be modified using the standardized prometheus.io annotations.</br>
+
+### Sample annotation
+```
+ annotations:
+   prometheus.io/path: "/path/to/metrics"
+   prometheus.io/port: "8080"
+   prometheus.io/scrape: "true"
+```
+Agent will only scrape if `prometheus.io/scrape` is set to `true`.</br> 
+The `prometheus.io/path` is optional, if not set, then it will default to `/metrics`. The rest of the configuration is set to default as well.
+
+## 2.2. prometheus scrape config json
+The configuration can be fine tuned by providing custom json in [prometheus-scrape-config.json](./prometheus-scrape-config.json). This exposes all available PrometheusEmitter parameters.</br>
+Json takes highest precedence and overrides other types (annotation and out of the box)</br>
+If a pod matches multiple configs, then the last matching config will be applied.
+
+### 2.2.1  Sample JSON with all supported parameters (including optional)
+```
+[
+    {
+        "podMatcher":
+        {
+            "namespaceRegex": "pod_namespace.*",
+            "podNameRegex": "sample-pod.*"
+        },
+        "config":
+        {
+            "path": "/path/to/metrics/endpoint",
+            "port": 9100,
+            "namespace": "push_metrics_to_namespace",
+            "allowMetrics": "sampleMetric1,sampleMetric2",
+            "compartmentId": "ocid1.compartment.oc1..sample",
+            "scheduleMins": 1
+        },
+        "disable": false
+    },
+    ...
+]
+```
+
+### 2.2.2 First class members
+| member | required | description |
+|--------|----------|-------------|
+| podMatcher | yes | Elements used to match pods |
+| config | no | Collection configuration for PrometheusEmitter data source of the matching pod. This is optional, if disable is set to `true` |
+| disable | no | This is optional and defaults to `false`. If set to `true`, then podMatcher is used to restrict matching pods from collecting PrometheusEmitter metrics |
+
+### 2.2.3 podMatcher
+| member | required | type | description |
+|--------|----------|----- | ------------|
+| namespaceRegex | yes | `string` | Complete regular expression to match pod's namespace |
+| podNameRegex | yes | `string`  | Complete regular expression to match pod's name |
+
+### 2.2.4 config
+| member | required | type | default | description |
+|--------|----------|----- | ------- | ----------- |
+| path | no | `string` | /metrics | Path on which metrics are being emitted, e.g. /metrics  |
+| port | yes | `int` | NA | Port on which metrics are being emitted |
+| namespace | no | `string` | pod_prometheus_emitters | OCI namespace to which metrics are pushed |
+| allowMetrics | no | `string` | * | Comma separated metrics allowed to be collected. Defaults to *, which means all |
+| compartmentId | no | `string` | Agent's compartmentId | Compartment to which metrics are pushed. If not provided, then metrics will be pushed to same compartment where agent is installed |
+| scheduleMins | no | `int` | 1 | Minute interval at which metrics are collected |

charts/mgmt-agent/resources/prometheus-scrape-config.json

@@ -0,0 +1,2 @@
+[
+]

charts/mgmt-agent/templates/env-configmap.yaml

@@ -15,3 +15,4 @@ data:
   {{- if .Values.deployment.cleanupEpochTime }}
   POD_CLEANUP_ID: "{{ .Values.deployment.cleanupEpochTime }}"
   {{- end }}
+  ENABLE_AUTOMATIC_PROMETHEUS_DETECTION: "{{ .Values.kubernetesCluster.enableAutomaticPrometheusDetection }}"

charts/mgmt-agent/templates/metrics-configmap.yaml

@@ -34,3 +34,5 @@ data:
     # list of comma separated metric names for Node (/api/v1/nodes/<node_name>/proxy/metrics/resource, /api/v1/nodes/<node_name>/proxy/metrics/cadvisor) metrics
     overrideAllowMetricsNode={{ .Values.kubernetesCluster.overrideAllowMetricsNode }}
     {{- end }}
+  prometheus-scrape-config.json: |
+{{ .Files.Get "resources/prometheus-scrape-config.json" | indent 4 }}

charts/mgmt-agent/values.schema.json

@@ -131,11 +131,16 @@
                           "type": "null"
                         }
                     ]
+                },
+                "enableAutomaticPrometheusDetection":
+                {
+                    "type": "boolean"
                 }
             },
             "required":
             [
-                "namespace"
+                "namespace",
+                "enableAutomaticPrometheusDetection"
             ]
         },
         "deployment":

charts/mgmt-agent/values.yaml

@@ -67,6 +67,9 @@ kubernetesCluster:
    overrideAllowMetricsKubelet:
    # -- Provide the specific list of comma separated metric names for Node (/api/v1/nodes/<node_name>/proxy/metrics/resource, /api/v1/nodes/<node_name>/proxy/metrics/cadvisor) metrics to be collected.
    overrideAllowMetricsNode:
+   # Setting this to true will automatically enable PrometheusEmitter metrics collection for eligible pods.
+   # To customize and get more information, refer ./resources/auto_prometheus_collection_readme.md
+   enableAutomaticPrometheusDetection: false
 
 deployment:
    security:

charts/oci-onm/values.yaml

@@ -56,6 +56,6 @@ oci-onm-mgmt-agent:
     # Follow steps documented at https://github.com/oracle/docker-images/tree/main/OracleManagementAgent to build docker image.
     image:
       # Replace this value with actual docker image URL for Management Agent
-      url: container-registry.oracle.com/oci_observability_management/oci-management-agent:1.7.0
+      url: container-registry.oracle.com/oci_observability_management/oci-management-agent:1.8.0
       # Image secrets to use for pulling container image (base64 encoded content of ~/.docker/config.json file)
       secret: