Add Telegraf/InfluxDB/Grafana compose stack for recording InfluxDB me…

…trics (#166)

* Add telegraf/influxdb/grafana compose stack

* Add commented-out defaults to system.env for telegraf metrics

* upgrade grafana version

* Bump Robomaker dependency to 5.2.2

* fix auto-provisioning of dashboard template

* Rename influx -> metrics

* fix initial dashboard datasource

* docker-compose -> docker compose

* Add telegraf env vars to eval and training docker compose files

* add docs for metrics

* tweak docs

* tweak docs

* Default to empty strings

* fix metric name in dashboard

---------

Co-authored-by: Matt Camp <matt@noise.net.nz>

bin/scripts_wrapper.sh

@@ -311,3 +311,11 @@ function dr-update-viewer {
   $DR_DIR/scripts/viewer/stop.sh "$@"
   $DR_DIR/scripts/viewer/start.sh "$@"
 }
+
+function dr-start-metrics {
+  $DR_DIR/scripts/metrics/start.sh "$@"
+}
+
+function dr-stop-metrics {
+  $DR_DIR/scripts/metrics/stop.sh "$@"
+}

defaults/dependencies.json

@@ -2,7 +2,7 @@
     "master_version": "5.2",
     "containers": {
         "rl_coach": "5.2.1",
-        "robomaker": "5.2.1",
+        "robomaker": "5.2.2",
         "sagemaker": "5.2.1"
     }
 }

defaults/template-system.env

@@ -26,4 +26,6 @@ DR_WEBVIEWER_PORT=8100
 # DR_DISPLAY=:99
 # DR_REMOTE_MINIO_URL=http://mynas:9000
 # DR_ROBOMAKER_CUDA_DEVICES=0
-# DR_SAGEMAKER_CUDA_DEVICES=0
+# DR_SAGEMAKER_CUDA_DEVICES=0
+# DR_TELEGRAF_HOST=telegraf
+# DR_TELEGRAF_PORT=8092

docker/docker-compose-eval.yml

@@ -31,4 +31,6 @@ services:
       - ROS_MASTER_URI=http://localhost:11311/
       - ROS_IP=127.0.0.1
       - GAZEBO_ARGS=${DR_GAZEBO_ARGS:-}
+      - TELEGRAF_HOST=${DR_TELEGRAF_HOST:-}
+      - TELEGRAF_PORT=${DR_TELEGRAF_PORT:-}
     init: true

docker/docker-compose-metrics.yml

@@ -0,0 +1,45 @@
+
+version: '3.7'
+
+networks:
+  default:
+    external: true
+    name: sagemaker-local
+
+services:
+  telegraf:
+    image: telegraf:1.18-alpine
+    volumes:
+    - ./metrics/telegraf/etc/telegraf.conf:/etc/telegraf/telegraf.conf:ro
+    depends_on:
+      - influxdb
+    links:
+      - influxdb
+    ports:
+    - '127.0.0.1:8125:8125/udp'
+    - '127.0.0.1:8092:8092/udp'
+
+  influxdb:
+    image: influxdb:1.8-alpine
+    env_file: ./metrics/configuration.env
+    ports:
+      - '127.0.0.1:8886:8086'
+    volumes:
+      - influxdb_data:/var/lib/influxdb
+
+  grafana:
+    image: grafana/grafana:10.4.2
+    depends_on:
+      - influxdb
+    env_file: ./metrics/configuration.env
+    links:
+      - influxdb
+    ports:
+      - '3000:3000'
+    volumes:
+      - grafana_data:/var/lib/grafana
+      - ./metrics/grafana/provisioning/:/etc/grafana/provisioning/
+
+volumes:
+  grafana_data: {}
+  influxdb_data: {}

docker/docker-compose-training.yml

@@ -45,4 +45,6 @@ services:
       - ROS_MASTER_URI=http://localhost:11311/
       - ROS_IP=127.0.0.1
       - GAZEBO_ARGS=${DR_GAZEBO_ARGS:-}
+      - TELEGRAF_HOST=${DR_TELEGRAF_HOST:-}
+      - TELEGRAF_PORT=${DR_TELEGRAF_PORT:-}
     init: true

docker/metrics/configuration.env

@@ -0,0 +1,9 @@
+# Grafana options
+GF_SECURITY_ADMIN_USER=admin
+GF_SECURITY_ADMIN_PASSWORD=admin
+GF_INSTALL_PLUGINS=
+
+# InfluxDB options
+INFLUXDB_DB=influx
+INFLUXDB_ADMIN_USER=admin
+INFLUXDB_ADMIN_PASSWORD=admin

docker/metrics/grafana/provisioning/dashboards/dashboard.yml

@@ -0,0 +1,7 @@
+apiVersion: 1
+
+providers:
+- name: 'Default'
+  folder: ''
+  options:
+    path: /etc/grafana/provisioning/dashboards