Sensors Unavailable / Repeatedly requesting tokens from Enphase - "Found empty token"

[dh99999]

I have an issue whereby it appears the integration is restarting during the processing of data.

It does not do it every time - it can work fine for 6 hours, but then fail 3 times in 10 minutes.

From the log I can see data being extracted from the envoy, but then as the processing should kick in, it appears to be reset and immediately request a new token from enphase. ie: It looks like it is crashing or otherwise restarting processing.

Simultaneously with this I see any template sensors that I have which use the enphase values receive an "unavailable" state for the sensors, which again makes me think the integration is restarting.

There are no stack traces that I can see in the home-assistant.log, or anything else that seems to indicate that the integration was restarted by HA, so at this point I am unsure how to identify the issue.

Is there another log I should be looking at to find this information?

2023-07-17 23:43:49.699 DEBUG (MainThread) [custom_components.enphase_envoy] Retrieved data from API: {'production': -9, 'daily_production': 12041, 'seven_days_production': 96181, 'lifetime_production': 14678023, ..... 2023-07-17 23:43:49.699 DEBUG (MainThread) [custom_components.enphase_envoy] Finished fetching envoy Envoy XXXXXXX data in 9.250 seconds (success: True)
... 2023-07-17 23:43:49.816 DEBUG (MainThread) [custom_components.enphase_envoy.envoy_reader] Checking Token value:
2023-07-17 23:43:49.816 DEBUG (MainThread) [custom_components.enphase_envoy.envoy_reader] Found empty token:
2023-07-17 23:43:54.091 DEBUG (MainThread) [custom_components.enphase_envoy.envoy_reader] Commissioned Token valid for 364.9999873635338 days

This may very well be an issue with my setup, but no other integrations seem to be impacted.

Any ideas?

[catsmanac]

Looks indeed like a restart of some kind. I see a scan took almost 10 seconds. Is that always the case?

To see more details in the log you can change logging level in configuration.yaml in the config folder by adding a logger section to display all information.

logger:
  default: info

But it will create a huge log file with lots of details.

[dh99999]

Looks indeed like a restart of some kind. I see a scan took almost 10 seconds. Is that always the case?

No, not always 10 seconds - it was generally around 5, with the call to home.json taking most of the time.

I am now running the "main" branch without the call to home.json and the scan is now taking less than 2 seconds.

I have also added some debugging around the calls to async_setup_entry and async_unload_entry to see if that shows anything associated with the integration being restarted.

I have not seen the issue since I have updated to the main branch about 24 hours ago - I will continue to monitor and report back.

I was previously on v0.0.13, but I was seeing the issue in v0.0.12 as well.

[posixx]

How is your envoy connected to the network? Through WiFi? If so, switch to an Ethernet connection as that will be much more stable and probably much faster..

[dh99999]

Envoy and HA are all connected via gigabit Ethernet... router and HA logs do not show any disconnects that I can see. Also not seeing any connection issues to other devices monitored by HA. Envoy is not reporting connectivity issues that I can see (though I must admit I don't have a good way to monitor this - I will setup a pingplotter if the issues recur)

I would have expected log entries from the integration if it was a connectivity issue?

Not sure if 5s was a long time for a scan to complete (3 phase, consumption + production CT metering, 17 micro inverters, no battery), but now that home.json isn't being called the scan is much faster.

[dh99999]

Closing.

Issue has not recurred since running the "main?" branch with home.json not being called.

Thanks for the help!