springfall2008
Predbat self heal - didn't heal from Websocket error
Describe the bug Looks like at 1809GMT the Web socket was closed and Predbat never recovered until I noticed and restarted Predbat at 2255GMT. The raspberry Pi5 was fully accessible during that time period as was the home assistant app connection.
Error: Web Socket exception in startup: Cannot connect to host
Expected behaviour I believe the predbat self heal mechanism should handle this?
Predbat version
Addon 1.2.4
Environment details
- Inverter and battery setup GivEnergy hybrid 3.6 with 9.5 & 5.2 batteries
- Standard HAOS installer or Docker Standard HAOS on pi5
- Anything else?
Screenshots If applicable, add screenshots to help explain your problem. The most useful ones can be your battery chart, the Predbat HTML plan and your current settings in HA.
Log file Attached
Predbat debug yaml file Not obtained but could use the files from my other ticket raised yesterday.
https://github.com/springfall2008/batpred/issues/2044
There are a few people that have reported this, its due to the access token expiring. The fix is to create your own HA access token for predbat to use. I've made this clearer in this documentation update https://github.com/gcoan/batpred/blob/main/docs/apps-yaml.md#home-assistant-connection
@gcoan i have an access token as per the documentation 😀. In the above log I've just REDACTED the token
`[09:07:22] INFO: Predbat init script running Running Predbat inside Add-on Your API key is: e1550e778REDACTED
Bootstrap Predbat Startup Add-on: Predbat Home Battery Prediction and Control
Add-on version: 1.2.4 You are running the latest version of this add-on.`
I do get this after the above line though
2025-02-26 09:07:23.309651: Warn: Failed to decode response <Response [404]> from http://192.168.251.77:8123/addons/self/info
2025-02-26 09:07:23.309651: Warn: Failed to decode response <Response [404]> from http://192.168.251.77:8123/addons/self/info
I am just doing final testing of a fix for this warning and will be pushing it through soon.
Its just a warning that the slug id can't be found, which only occurs if you have set ha_url. The only impact is some of the log messages, so its not a major problem, but I'm fixing it
As for why you get the websocket error when you have a ha_key set, I'm sorry, no idea.
Is there nothing else that predbat can do to restart itself or am I looking into HA automation to try to fix Predbat?
I did just push (yesterday) a new release with an auto-restart for some other conditions, maybe worth a try
@springfall2008 Many thanks, I'll have a look. Is there anything else I can do regarding the HA_Key and why I get these web socket errors when the HA_Key is present?
Battery didn't charge last night
2025-03-16 18:02:41.310848: Warn: Web Socket closed, will try to reconnect in 5 seconds - error count 3 2025-03-16 18:02:46.312296: Info: Start socket for url HTTP:// REDACTED:8123/api/websocket 2025-03-16 18:02:46.314145: Error: Web Socket exception in startup: Cannot connect to host REDACTED:8123 ssl:default [Connect call failed (REDACTED', 8123)]
I suspect this is my root cause problem
https://github.com/home-assistant/core/issues/135428
OK might have noticed something when I trigger HA to restart from Settings > System > power button > Restart Home Assistant (yellow icon).
Predbat never seems to recovery itself via this method of HA restart, gives me the Web Socket errors and needs the addon to be restarted.
2025-03-17 15:05:02.082611: Warn: Historical day 7 has 55 minutes of gap in the data, filled from 24.98 kWh to make new average 25.98 kWh (percent 96%) 2025-03-17 15:05:04.068475: Info: record_status Demand 2025-03-17 15:09:18.572912: Warn: Web Socket closed, will try to reconnect in 5 seconds - error count 0 2025-03-17 15:09:23.574704: Info: Start socket for url http://192.168.251.77:8123/api/websocket 2025-03-17 15:09:23.585074: Error: Web Socket exception in startup: Cannot connect to host 192.168.251.77:8123 ssl:default [Connect call failed ('192.168.251.77', 8123)]
OK might have noticed something when I trigger HA to restart from Settings > System > power button > Restart Home Assistant (yellow icon).
Predbat never seems to recovery itself via this method of HA restart, gives me the Web Socket errors and needs the addon to be restarted.
I'm not surprised this can cause problems because you're restarting HA so the socket connection to HA gets broken. If you restart via advanced options / restart HA and all add-ons then it will be OK.
Whether Predbat can trap these errors is a different question
I'll cve that with the predbat experts to set off its doable 😋
For now I've created a HA automation using these triggers
`triggers:
- trigger: state
entity_id:
- switch.predbat_active for: hours: 0 minutes: 20 seconds: 0 to: null
- trigger: state
entity_id:
- predbat.status to: "Error: Exception raised Auto-restart triggered" for: hours: 0 minutes: 20 seconds: 0
- trigger: state
entity_id:
- predbat.status to: >- "Inverter 0 read bad REST data from HTTP:// REDACTED HA IP:6345/runAll - REST will be disabled" for: hours: 0 minutes: 20 seconds: 0`
There is a predbat error monitor I wrote in the documentation that you could use https://springfall2008.github.io/batpred/output-data/#predbat-error-monitor, it checks for a number of other indicators of predbat being stuck
switch.predbat_active being stuck on for a long period of time makes sense, I'll add that
the error: exception, I have that, but a more broader check for the text 'error'
bad REST data, I don't check for that, possibly could, but I hardly ever see that one
anyway, have a look at the automation
@gcoan now thats sweet, didn't spot that part of the page before. Now adopted into my automation