Aqara Pet Feeder sometimes loses its schedule when Z2M is restarted

[rccoleman]

What happened?

I have 3 separate daily feeding times configured in the Aqara Pet Feeder (ZNCWWSQ01LM) through Z2M, and it seems like it sometimes loses its local schedule when I restart Z2M. This was also reported here: https://community.home-assistant.io/t/aqara-pet-feeder-in-z2mqtt-wrong-schedule-after-reboot/689354

As I wrote there, sometimes the schedule disappears in the Z2M UI, and then the schedule reappears if I switch between tabs and reload the page. I suspect that there may be a race condition that causes the initial configuration that Z2M performs at startup to sometimes send an empty schedule even when Z2M does appear to have the proper schedule based on the UI.

It's now in a state where I can't get it to feed automatically based on a schedule that I set regardless of whether I change or add a time and hit apply or force a reconfiguration. It just seems completely unreliable.

I can reliably trigger a feeding from the exposed "Feed" entity in HA, so I've now created an automation to ensure that meals are dispensed, but I can run any experiments that might help.

What did you expect to happen?

I expect the device to feed based on the schedule that I set in Z2M.

How to reproduce it (minimal and precise)

It's hard to say exactly what causes it, but it definitely seems related to restarting Z2M.

Zigbee2MQTT version

1.36.0-dev commit: 38ed953

Adapter firmware version

20221226

Adapter

Tubeszb USB ZStack adapter

Setup

Container installation

Debug log

log.txt

[Klikini]

This also happens to me, but I’m using a ConBee II.

[rccoleman]

I wonder if it's related to nurikk/zigbee2mqtt-frontend#1939, which calls out a mismatch between the frontend and backend. I found that while trying to debug this issue and I noticed that I couldn't find the manuSpecificLumi cluster in the Dev Console for the Pet Feeder, but it was clearly listed in the cluster info tab.