Two of my ESPHome devices keep disconnecting from HA API

notsphinx commented 1 day ago

The problem

I have 103 ESPHome devices on my home network. Two of the devices disconnect from Home Assistant API. They are connected fine on WiFi and I can manage them using the web interface, but none of their related automations work. I can re-add the device to HA after which they stay connected for about 2 minutes and then drop off from HA even though they are reachable on the network. This is always specific to two of these devices (different types, so different yaml files) and this problem persists across device reboots and HA reboots.

I have multiple devices of each of these types, so not specific to one particular device type. The WiFi signal strength according to these devices is about -57 or -58 dBm. I do not believe this is WiFi related since that could cause some degree of randomness.

Which version of ESPHome has the issue?

2024.11.2

What type of installation are you using?

pip

Which version of Home Assistant has the issue?

2024.11.3

What platform are you using?

BK72XX

Board

Cree BR30 Light bulb (generic-bk7231t-qfn32-tuya)

Component causing the issue

api

YAML Config

substitutions:
  name: cree-br30
  friendly_name: Cree BR30

esphome:
  name: ${name}
  name_add_mac_suffix: true

bk72xx:
  board: generic-bk7231t-qfn32-tuya
  framework:
    version: recommended

preferences:
  flash_write_interval: 1min

<<: !include common/network-lite.yaml

# Sync time with Home Assistant.
time:
  - platform: homeassistant
    id: homeassistant_time

button:
- platform: restart
  id: restart_button
  name: Restart
- platform: safe_mode
  id: safe_mode_restart_button
  name: Safe Mode Restart

debug:
  update_interval: 60s

# Individual on/off sensors
binary_sensor:
# Reports if this device is Connected or not
- platform: status
  name: Status

# Individual data sensors
sensor:
# Reports the WiFi signal strength/RSSI in dB, see https://esphome.io/components/sensor/wifi_signal.html
- platform: wifi_signal
  name: Signal
  update_interval: 60s

Anything in the logs that might be useful for us?

Snapshot of logs between the connect and disconnect events:

[15:38:43][W][api.connection:092]: : Socket operation failed: BAD_INDICATOR errno=9
[15:38:43][D][api:103]: Accepted 172.x.x.x
[15:38:43][W][api.connection:092]: : Socket operation failed: HANDSHAKESTATE_READ_FAILED errno=9
[15:38:47][D][api:103]: Accepted 172.x.x.x
[15:38:47][W][component:237]: Component api took a long time for an operation (126 ms).
[15:38:47][W][component:238]: Components should block for at most 30 ms.
[15:38:47][D][api.connection:199]: aioesphomeapi (172.x.x.x) requested disconnected
[15:38:47][D][api:103]: Accepted 172.x.x.x
[15:38:47][W][component:237]: Component api took a long time for an operation (122 ms).
[15:38:47][W][component:238]: Components should block for at most 30 ms.
[15:38:47][D][api.connection:1446]: Home Assistant 2024.11.3 (172.x.x.x): Connected successfully
[15:38:47][D][time:050]: Synchronized time: 2024-12-02 15:38:47
[15:38:54][D][sensor:093]: 'Signal': Sending state -59.00000 dBm with 0 decimals of accuracy
[15:39:00][W][component:237]: Component wifi took a long time for an operation (116 ms).
[15:39:00][W][component:238]: Components should block for at most 30 ms.
[15:39:28][D][sensor:093]: 'Uptime': Sending state 1594.68616 minutes with 0 decimals of accuracy
[15:39:41][D][sensor:093]: 'Signal Percent': Sending state 84.00000 Signal % with 0 decimals of accuracy
[15:39:54][D][sensor:093]: 'Signal': Sending state -59.00000 dBm with 0 decimals of accuracy
[15:40:28][D][sensor:093]: 'Uptime': Sending state 1595.68616 minutes with 0 decimals of accuracy
[15:40:41][D][sensor:093]: 'Signal Percent': Sending state 84.00000 Signal % with 0 decimals of accuracy
[15:40:54][D][sensor:093]: 'Signal': Sending state -58.00000 dBm with 0 decimals of accuracy
[15:41:28][D][sensor:093]: 'Uptime': Sending state 1596.68616 minutes with 0 decimals of accuracy
[15:41:41][D][sensor:093]: 'Signal Percent': Sending state 84.00000 Signal % with 0 decimals of accuracy
[15:41:54][D][sensor:093]: 'Signal': Sending state -60.00000 dBm with 0 decimals of accuracy
[15:42:28][D][sensor:093]: 'Uptime': Sending state 1597.68616 minutes with 0 decimals of accuracy
[15:42:41][D][sensor:093]: 'Signal Percent': Sending state 82.00000 Signal % with 0 decimals of accuracy
[15:42:54][D][sensor:093]: 'Signal': Sending state -59.00000 dBm with 0 decimals of accuracy
[15:43:02][W][component:237]: Component wifi_info.text_sensor took a long time for an operation (54 ms).
[15:43:02][W][component:238]: Components should block for at most 30 ms.
[15:43:28][D][sensor:093]: 'Uptime': Sending state 1598.68616 minutes with 0 decimals of accuracy
[15:43:41][D][sensor:093]: 'Signal Percent': Sending state 82.00000 Signal % with 0 decimals of accuracy
[15:43:54][D][sensor:093]: 'Signal': Sending state -60.00000 dBm with 0 decimals of accuracy
[15:44:28][D][sensor:093]: 'Uptime': Sending state 1599.68616 minutes with 0 decimals of accuracy
[15:44:41][D][sensor:093]: 'Signal Percent': Sending state 82.00000 Signal % with 0 decimals of accuracy
[15:44:54][D][sensor:093]: 'Signal': Sending state -58.00000 dBm with 0 decimals of accuracy
[15:45:28][D][sensor:093]: 'Uptime': Sending state 1600.68616 minutes with 0 decimals of accuracy
[15:45:41][D][sensor:093]: 'Signal Percent': Sending state 86.00000 Signal % with 0 decimals of accuracy
[15:45:54][D][sensor:093]: 'Signal': Sending state -56.00000 dBm with 0 decimals of accuracy
[15:46:28][D][sensor:093]: 'Uptime': Sending state 1601.68616 minutes with 0 decimals of accuracy
[15:46:41][D][sensor:093]: 'Signal Percent': Sending state 88.00000 Signal % with 0 decimals of accuracy
[15:46:54][D][sensor:093]: 'Signal': Sending state -55.00000 dBm with 0 decimals of accuracy
[15:47:09][D][time:050]: Synchronized time: 2024-12-02 15:47:09
[15:47:09][D][time:050]: Synchronized time: 2024-12-02 15:47:09
[15:47:23][D][api.connection:199]: Home Assistant 2024.11.3 (172.x.x.x) requested disconnected

There are no logs at the default log level on HA when the disconnect happens.

Additional information

No response

ssieb commented 22 hours ago

The disconnect is coming from the HA side, not the device. There are no logs on HA about this? Check both the general ones and the device specific ones.

How are you getting these logs with the mac suffix enabled?

notsphinx commented 13 hours ago

Mystery partially resolved:

I looked in the HA logs with the device name instead of the device IP address. It looks like HA is attempting to connect to an IP address that is different than what the DHCP server assigned to the device. I tested by configuring the DHCP server to assign the address that HA was trying to connect to and the ESPHome device's connection to HA API is stable.

Where is the IP address of the device cached in HA? And why doesn't it auto update if the IP address changes?

esphome / issues