I wanted to use the Service Info sensor if an App is running (in particular Zigbee2MQTT). Today I noticed that the Service Info (proviced by supervisor) is staying at the state "running" for several minutes before it changes to something else (eg unavailable). Well if at all.
If this is normal behavior then the sensor is completely useless.
Am I doing something wrong or is this intended?
Yes, I mean that one. Mine keeps staying at "Running" for several minutes even though the App is not running anymore.
Yep, you're right. Had a closer look at my logs & the state of the sensor is off by 2-3 mins from when the outage occurred.
No idea why, though I suspect it's either polling or could be waiting for the absence/presence of the health message, such as this one:
[2026-06-20 10:53:36] info: z2m:mqtt: MQTT publish: topic 'zigbee2mqtt/bridge/health', payload '{"response_time":1781945616334,"os":{"load_average":[0.11,0.05,0.01],"memory_used_mb":1144.33,"memory_percent":29.2238},"process":{"uptime_sec":142220,"memory_used_mb":139.66,"memory_percent":3.5665},"mqtt":{"connected":true,"queued":0,"published":21085,"received":297},
If that's the case, then the health message is fired every 10 minutes, which might explain the delay.
Either way, a couple of minutes here or there doesn't affect me. As long as I'm suitably informed when my network is down, I'm fine with that. It's not like I'm running something mission critical.
Well, with security related devices 10 minutes (even 5 minutes) is a very long time.
As a workaround I am using the state of button.zigbee2mqtt_bridge_restart. If it is unavailable then Zigbee2MQTT is off.
But there are other Apps (eg NodeRed) where I have not yet found a workaround.