ConsumptionX timeout on one device makes all its entities unavailable, including the switch

[ValentinStrach]

Describe the bug

A ConsumptionX timeout on a single device makes all entities of
that device unavailable, including the switch — even though ToggleX
and Electricity work fine on the same device.

I have two MSS305 plugs that are identical in every way: same hardware
revision (HW 8.1.0), same firmware (FW 8.3.15), same Meross account,
same Wi-Fi network, same MQTT broker.

• Plug A ("Prise clôture"): works perfectly. Switch controllable,
  power/voltage/current and energy counter all reporting.
• Plug B ("Chargeurs Power For All"): Appliance.Control.ConsumptionX
  times out (10 s) on every polling cycle. As a result,
  switch.*_main_channel and sensor.*_energy_kwh_main_channel are
  permanently unavailable, while sensor.*_power_w_main_channel,
  *_voltage_v_* and *_current_a_* keep updating normally every
  ~30 seconds.

So the device is clearly online and reachable — only one namespace
fails.

Key point: ToggleX is not the problem. I can turn plug B on and
off without any issue from the Meross Android app. Only Home Assistant
shows the switch as unavailable, because the failing ConsumptionX
call aborts the whole async_update() and every entity that depends on
it reports unavailable.

Expected behaviour

A timeout on one namespace should not invalidate entities belonging to
other, working namespaces. Ideally async_update() should tolerate
partial failures: log the failing namespace and keep serving the
namespaces that responded.

Things I already tried, none of which helped

• Reloading the integration
• Full Home Assistant restart
• Disabling sensor.*_energy_kwh_main_channel in the entity registry —
  the ConsumptionX request is still issued, so there is no way to
  work around this from the HA side

Your environment

HomeAssistant version: -- 2026.8.1 (Core) --
Hassio Version (if applicable): -- Supervisor 2026.07.5, OS 18.2 --
Hardware environment: -- Home Assistant Green --

Integration: meross_cloud (HACS), low level library 0.4.10.4
Device: MSS305, HW 8.1.0, FW 8.3.15, class mss305:8.1.0:8.3.15
MQTT host: mqtt-eu-5.meross.com:443

Logs taken when the issue happened

ERROR (meross_iot.manager) Timeout occurred while waiting a response
for message b'{"header":{"from":"/app/4414882-xxxx/subscribe",
"messageId":"f12c8948c42c912d783b2d97a124e010","method":"GET",
"namespace":"Appliance.Control.ConsumptionX","payloadVersion":1,
"sign":"81e0802dffba2a81137ef38f9731c198","timestamp":1786532225,
"triggerSrc":"Android","uuid":"240904419797245100000000000a5555"},
"payload":{"channel":0}}' sent to device uuid
240904419797245100000000000a5555. Timeout was: 10.000000 seconds.
Mqtt Host: mqtt-eu-5.meross.com:443.

ERROR (homeassistant.components.switch) meross_cloud: Error on device update!
ERROR (homeassistant.components.sensor) meross_cloud: Error on device update!

WARNING (meross_iot.manager) Found a new device Chargeurs Power For All
(mss305, HW 8.1.0, FW 8.3.15, class: mss305:8.1.0:8.3.15) that has
become online while we were offline.

WARNING (homeassistant.components.switch) Setup of switch platform
meross_cloud is taking over 10 seconds.
WARNING (homeassistant.components.sensor) Setup of sensor platform
meross_cloud is taking over 10 seconds.

[donations-bot-prod]

Donation bot here. Supporters can mark their interest in this issue by adding a comment
starting with "+1" (without quotes) or adding the "#supporter_sponsor" keyword
(without quotes) in their comments.

You can become a donor by subscribing as Sponsor from the GitHub sponsorship page or from the BuyMeACoffee page.

Issues marked by donors will be labeled accordingly and handled with special care