[Question]: Home Assistant automation using MQTT events; no event id sometimes?

[esand]

What is your question:

Is there a the possibility that Frigate would send an MQTT event without an "after" state and/or an event id?

My Home Assistant automation starts like this:

alias: Security - Person at front door
description: ""
trigger:
  - platform: mqtt
    topic: frigate/events
    payload: Front_doorbell/new
    value_template: "{{ value_json.after.camera }}/{{ value_json.type }}"
    variables:
      id: >-
        {{ trigger.payload_json.after.id if trigger is defined and
        trigger.payload_json is defined else '' }}

and everything seems to work rather well, but every once in a while, I get these errors in HA logs: Template variable warning: 'id' is undefined when rendering '{{ id }}'. To make it more difficult to debug, due to the error, HA doesn't log an event entry so I can't check the JSON payload to see what's going on, so I'm totally in the dark.

This is the first place I try to use {{ id }} in my automation - the beginning of my action sequence:

action:
  - alias: Wait for a Person on the Doorstep
    repeat:
      sequence:
        - alias: Wait for an uptate to the Frigate event
          wait_for_trigger:
            - platform: mqtt
              topic: frigate/events
              payload: "{{ id }}"
              value_template: "{{ value_json.after.id }}"
              alias: When an update to the Frigate event is received
          timeout:
            hours: 0
            minutes: 1
            seconds: 0
            milliseconds: 0
          continue_on_timeout: false

If it matters, the automation is set to parallel mode. Any ideas why I'd be getting those errors saying id is undefined?

[esand]

It took a few days but it happened again (seems less frequent now), however this time it seems to have happened when there were no recorded Frigate events at all. I went in to Frigate to look for any events around the time of the error and there's nothing. I have some full-time recording from the camera in question and exactly when the error occurred and nothing that seems like it could trigger an event happened (plus, no actual recorded event).

It now seems slightly more plausible that either Home Assistant triggered the automation itself, or MQTT sent some kind of ghost event? Is there any remote chance that Frigate could send an MQTT event without recording an event?

[NickM-27]

You should look at the automation trace and see what triggered the automation. no one else has reported anything like this

[esand]

That's the kicker - there is no trace of the event ever firing. I even increased the number of traces so I could confirm and sure enough, I have events before and after, but at the moment that error fires, nothing. In this latest error, it's not even close to another event so it can't even be my wait timeout (of 1 minute).

I'm starting to think it may not be Frigate for the reasons in my previous post, plus there seems to be nothing in the Frigate, go2rtc or nginx logs for the time of the error either. And yeah, I seem to be on my own on this one. Have searched high and low and can't find anyone else reporting a similar issue.

Should Home Assistant show a trace event entry even if there was an error in the event configuration? That would at least rule out Frigate and MQTT for good if it should log a trace.

[esand]

I'm going to close this issue, as the error hasn't shown up for a few days now and there's been ton of events that should have triggered it if it was related to my automation in any way. I have a feeling it was something in Home Assistant that a restart of HA or full restart of my container cleared out.

[esand]

For posterity, it appears the issue may in fact be due to using wait_for_trigger in the automation.

I modified my automation and added a variables section to the wait to re-define the id variable in case it was getting lost, and my logs captured a bunch of entries saying that wait was not defined. This means that the wait_for_trigger somehow caught an event, but didn't define the wait variable.

[Welsyntoffie]

Think I have the same issue. Recently moved to home assistant core on docker.
Strange is, that this works now and then, but mostly not. Used to work perfect on hassos.

Template variable warning: 'event_id' is undefined when rendering '{{event_id}}'

config.yml
shell_command: last_detected_snap: python3 /config/bashscript/last_detect_snap.py "{{ event_id }}"

py script

import datetime
import io
import json
import requests
import sys
from PIL import Image
FRIGATE_HOST = "http:https://172.22.2.111:5005"
event_id = sys.argv[1]
json = requests.get(f"{FRIGATE_HOST}/api/events/{event_id}").json()
img_bytes = requests.get(f"{FRIGATE_HOST}/api/events/{event_id}/snapshot.jpg?bbox=1").content
img = Image.open(io.BytesIO(img_bytes))
img.save("/config/www/cloud_plate/camera_image/last_detected_p.jpg")

Lastly this error comes up.
Error running command: python3 /config/bashscript/last_detect_snap.py "{{ event_id }}", return code: 1

[esand]

I've modified my original automation a bit, but the core component of creating an id variable and using it in the exact same locations as I originally posted remain the same. I don't get the error nearly as often as I used to, but it still pops up once in a while.

Most recently, I checked Frigate's events and right around the time I had the {{ id }} error, I had an event for one of my cameras but it was not my Front_doorbell camera - that one had no recorded events for that time. It made me wonder if there's something happening with my repeat/wait loop, however I do have a 1m timeout specified.

My only guess right now is that although I issue a Stop when my automation triggers a notification to exit everything, the wait might be lingering, and eventually another Frigate event happens, and causes it to test the payload of {{ id }} which has now been wiped since the timeout passed, resulting in the error.

Are you running your python script within a loop of some sort?

[Welsyntoffie]

No loop. I cant even remember where I got this snippet of code. This was my automation,
I have also deleted this and made an automation per camera and a python script per camera. Gives me the same results without having to wary about arguments and variables passed to the python script.
I have checked the file permissions, deleted and recreated the files, renamed the files. Yesterday I gave up.

alias: Last_detection_snap
description: ""
trigger:
  - platform: mqtt
    topic: frigate_mint/events
    variables:
      camera: "{{ trigger.payload_json[\"after\"][\"camera\"] }}"
      id: "{{ trigger.payload_json[\"after\"][\"id\"] }}"
condition:
  - condition: template
    enabled: true
    value_template: "{{trigger.payload_json[\"type\"] == \"new\" }}"
  - condition: template
    enabled: true
    value_template: "{{trigger.payload_json[\"after\"][\"label\"] == \"person\" }}"
action:
  - service: shell_command.last_detected_snap
    data_template:
      event_id: "{{event_id}}"
trace:
  stored_traces: 20
mode: queued
max: 3