bi-directional use?

[Flipsoo]

Hello, Ive tried to get a setup running where 2 instances of hassio connect together, but while i have to restart one instance to get the custom component running, the second instance only has the entities but doesnt get new state changes - the one last rebooted works great but the other doesnt. is there a trick or something in the readme im missing for that behavior?

Greetings to all of you:)

[marc-romu]

I've tried to connect two instances of home assistant together with remote_homeassistant version 3.1. My configuration is something like this:

Instance 1

remote_homeassistant:
  instances:
  - host: 192.168.10.1
    port: 8123
    entity_prefix: "ha2_"
    access_token: !secret access_token_to_2
    verify_ssl: false
    secure: false
    include:
      entities:
      - sensor.indoor_temperature
      - sensor.indoor_humidity
      - sensor.indoor_co2
      - light.bedroom
    subscribe_events:
      - state_changed
      - service_registered
      - mobile_app_notification_action

Instance 2

remote_homeassistant:
  instances:
  - host: 192.168.20.1
    port: 8123
    entity_prefix: "ha1_"
    access_token: !secret access_token_to_1
    verify_ssl: false
    secure: false
    include:
      entities:
      - sensor.downstairs_temperature
      - sensor.upstairs_temperature

Problem

While one instance is stopped or being restarted, the other one works fine. When both instances are running, UI heavily decreases performance and it's almost impossible to interact with. I haven't been able to see what happens in live because both instances get collapsed and I cannot see any relevant error or warning in logs. I think it could be happening something like this:

• Instance 1 updates some sensor
• Instance 2 is listening to event state_changed on instance 1, so it updates sensors
• Instance 1 is listening to event state_changed on instance 2, so it updates sensors
• Instance 2 is listening to event state_changed on instance 1, so it updates sensors
• (and it gets stuck in a loop)

I've tried to disable subscription to state_changed and service_registered on instance 1 and now there is no performance decrease, but sensors are not updated either.

    subscribe_events:
#      - state_changed
#      - service_registered
      - mobile_app_notification_action

Is there any known solution for it? Thanks!

[mindtripper]

@marc-romu I have the same issue using any version above 2.3. CPU goes through the roof on both instances. Using the 2.3 on one instance and 3.3 on the other works. Same goes for 2.3 on both instances. No performance impact.

Would like to be able to run 3.3 on both instances.

I only use the following;

    subscribe_events:
    - state_changed

[postlund]

I removed the service_registered event internally so it should work as before. A new release is out, should be available via HACS soon.

[mindtripper]

Hah! Problem solved. Running 3.4 on both instances with normal system utilization.

Thank you!

[marc-romu]

Yes, now works! Thanks!