Refactor and extend prometheus component

[jzucker2]

Component name

prometheus

Link to component documentation on our website

https://esphome.io/components/prometheus/

Describe the enhancement

I'm going to start with one big discussion around this component, as I am a big fan and a heavy user. We can break out ideas if that's considered best. I will let someone else make that call.

• Add remaining missing components (event, text, button, alarm control panel, ?)
• Provide a basic example grafana dashboard
• Some basic alerts as well
• Update docs ([includes some example prometheus metrics)
• Reduce code duplication
• Optionally allow more fine-grained control over sensors/components to include or exclude
• Create a v2 metrics pattern that has less distinct metric names and more dynamic labels.

Current component allows relabeling and a boolean for internal components.

But what if there was something like:

prometheus:
  include:
    entities:
      - some_important_sensor_id
    components:
      - sensor
      - binary_sensor
  exclude:
    entities:
      - some_noisy_trivial_sensor_id
    components:
      - datetime

This might more like having only an explicit include or exclude and both might not be required. But it could help reduce prometheus payload size, especially if there's a handful of sensors or a specific component not needing metrics.

As well, I think it'd be interesting to consider a v2 of the metrics that has an output that is simpler with more labels.

So instead of this as a current example:

#TYPE esphome_sensor_value gauge
#TYPE esphome_sensor_failed gauge
esphome_sensor_failed{id="dev_idf_prometheus_goober_wifi_signal_db",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober WiFi Signal dB"} 0
esphome_sensor_value{id="dev_idf_prometheus_goober_wifi_signal_db",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober WiFi Signal dB",unit="dBm"} -35
esphome_sensor_failed{id="dev_idf_prometheus_goober_wifi_signal_percent",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober WiFi Signal Percent"} 0
esphome_sensor_value{id="dev_idf_prometheus_goober_wifi_signal_percent",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober WiFi Signal Percent",unit="Signal %"} 100
#TYPE esphome_binary_sensor_value gauge
#TYPE esphome_binary_sensor_failed gauge
esphome_binary_sensor_failed{id="dev_idf_prometheus_goober_button",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober Button"} 0
esphome_binary_sensor_value{id="dev_idf_prometheus_goober_button",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober Button"} 0.000000
#TYPE esphome_number_value gauge
#TYPE esphome_number_failed gauge
esphome_number_failed{id="template_number",area="Dev",node="devidfprometheus",name="Template number"} 0
esphome_number_value{id="template_number",area="Dev",node="devidfprometheus",name="Template number"} 0.000000
#TYPE esphome_select_value gauge
#TYPE esphome_select_failed gauge
esphome_select_failed{id="template_select",area="Dev",node="devidfprometheus",name="Template select"} 0
esphome_select_value{id="template_select",area="Dev",node="devidfprometheus",name="Template select",value="two"} 1.0
#TYPE esphome_update_entity_state gauge
#TYPE esphome_update_entity_info gauge
#TYPE esphome_update_entity_failed gauge
esphome_update_entity_failed{id="firmware_update",area="Dev",node="devidfprometheus",name="Firmware Update"} 1
#TYPE esphome_text_value gauge
#TYPE esphome_text_failed gauge
esphome_text_failed{id="template_text",area="Dev",node="devidfprometheus",name="Template text"} 0
esphome_text_value{id="template_text",area="Dev",node="devidfprometheus",name="Template text",value="Hello World"} 1.0
#TYPE esphome_event_value gauge
#TYPE esphome_event_failed gauge
esphome_event_failed{id="template_event",area="Dev",node="devidfprometheus",name="Template Event"} 0
esphome_event_value{id="template_event",area="Dev",node="devidfprometheus",name="Template Event",last_event_type="custom_event_1"} 1.0

We have something more like:

#TYPE esphome_component_value gauge
#TYPE esphome_component_failed gauge
esphome_component_failed{component_type="sensor",id="dev_idf_prometheus_goober_wifi_signal_db",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober WiFi Signal dB"} 0
esphome_component_value{component_type="sensor",id="dev_idf_prometheus_goober_wifi_signal_db",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober WiFi Signal dB",unit="dBm"} -35
esphome_component_failed{component_type="sensor",id="dev_idf_prometheus_goober_wifi_signal_percent",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober WiFi Signal Percent"} 0
esphome_component_value{component_type="sensor",id="dev_idf_prometheus_goober_wifi_signal_percent",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober WiFi Signal Percent",unit="Signal %"} 100
esphome_component_failed{component_type="binary_sensor",id="dev_idf_prometheus_goober_button",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober Button"} 0
esphome_component_value{component_type="binary_sensor",id="dev_idf_prometheus_goober_button",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober Button"} 0.000000
esphome_component_failed{component_type="binary_sensor",id="dev_idf_prometheus_goober_connected_status",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober Connected Status"} 0
esphome_component_value{component_type="binary_sensor",id="dev_idf_prometheus_goober_connected_status",area="Dev",node="devidfprometheus",name="Dev IDF Prometheus Goober Connected Status"} 0.000000
esphome_component_failed{component_type="number",id="template_number",area="Dev",node="devidfprometheus",name="Template number"} 0
esphome_component_value{component_type="number",id="template_number",area="Dev",node="devidfprometheus",name="Template number"} 0.000000
esphome_component_failed{component_type="select",id="template_select",area="Dev",node="devidfprometheus",name="Template select"} 0
esphome_component_value{component_type="select",id="template_select",area="Dev",node="devidfprometheus",name="Template select",value="two"} 1.0
#TYPE esphome_update_entity_state gauge
#TYPE esphome_update_entity_info gauge
esphome_component_failed{component_type="update",id="firmware_update",area="Dev",node="devidfprometheus",name="Firmware Update"} 1
#TYPE esphome_text_value gauge
esphome_component_failed{component_type="text",id="template_text",area="Dev",node="devidfprometheus",name="Template text"} 0
esphome_component_value{component_type="text",id="template_text",area="Dev",node="devidfprometheus",name="Template text",value="Hello World"} 1.0
#TYPE esphome_event_value gauge
esphome_component_failed{component_type="event",id="template_event",area="Dev",node="devidfprometheus",name="Template Event"} 0
esphome_component_value{component_type="event",id="template_event",area="Dev",node="devidfprometheus",name="Template Event",last_event_type="custom_event_1"} 1.0

So this would remove some unnecessary code paths for current metrics that have failed metric code but can't reach it. It also solved the problem of needing 2 new metric types for each component left to add (and currently added, as well).

This would likely need to be done with a v1/v2 flag of some kind to not break old set ups and give time to switch over. But this would greatly simplify dashboards and alerting.

This would also sync up nicely with the work in in the include/exclude lists.

Some of this is already in motion:

• [prometheus] Add event and text base components metrics esphome#10240
• [prometheus] Add prometheus datetime metrics esphome#9941
• [prometheus] Avoid generating unused light color metrics to reduce memory usage on ESP8266 esphome#9530 (this is not my pr, just worth mentioning as an efficiency improvement)

Use cases

This is used for grafana dashboards, alerting, and debugging.

I think these changes/improvements could help with easy and generic solutions to all of the above, but especially the 3 main use cases listed above.

Anything else?

No response

[github-actions[bot]]

🏷️ I've automatically added the component: prometheus label to help categorize this feature request.