search

Koenkk / zigbee2mqtt

Zigbee 🐝 to MQTT bridge 🌉, get rid of your proprietary Zigbee bridges 🔨
https://www.zigbee2mqtt.io
GNU General Public License v3.0
11.95k stars 1.66k forks source link

Disable processing for TRADFRI wireless dimmer (ICTC-G-1) #3865

Closed vdkeybus closed 4 years ago

vdkeybus commented 4 years ago

Help for debugging

What happened

The Ikea ICTC-G dimmers seem to have processing built in to Z2M. There must be some bug that causes a command to be repeated to the backend (e.g. HA) indefinitely and bringing the whole system down.

At least the Zigbee sniffer indicates that the dimmers aren't sending any commands at the time Z2M is.

Question

I would like to debug this, but I'm unfamiliar with the Z2M code. Can someone point me to the right directory / source file ? I also see that there's some settings possible, but I'd like to use the auto-detect if possible.

Thanks

Koenkk commented 4 years ago

Please try disabling the legacy integration, the new integration only provides the actual commands being send (https://www.zigbee2mqtt.io/devices/ICTC-G-1.html#legacy-integration)

vdkeybus commented 4 years ago

Thanks for the quick response.

I saw the option, but I wanted to see if I could fix the bizarre error; it has brought my system down countless times, and even for 'legacy' users it's a bug. Apart from that, I want to avoid relying on config files with exceptions as much as possible.

I'm wondering if this option shouldn't be turned off by default. Or even removed.

It looks like the integration tries to complete missing messages from the remote, to either smoothen the transition or terminate every sequence with 'rotate_stop' when that last command isn't received. When the bug happens, 'rotate_stop' is never issued and 'rotate_left' / 'rotate_right' is repeated ad infinitum. (For some reason, that completely bogs down the HA server, which is a bug in itself too.)

For a z2m client, there is no difference between passing the original message sequence, which could possibly not contain the 'stop' message, and the infinite sequence that doesn't contain the 'stop' either. So it wouldn't matter to them if legacy would be turned off by default ?

Op di 7 jul. 2020 om 13:58 schreef Koen Kanters notifications@github.com:

Please try disabling the legacy integration, the new integration only provides the actual commands being send ( https://www.zigbee2mqtt.io/devices/ICTC-G-1.html#legacy-integration)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Koenkk/zigbee2mqtt/issues/3865#issuecomment-654806129, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSNXH2OEEVEIPFY2R77CT3R2MEXBANCNFSM4OSXGOLQ .

Koenkk commented 4 years ago

I'm wondering if this option shouldn't be turned off by default. Or even removed.

It will be removed in zigbee2mqtt 2.0.0, but as this is a breaking change it is kept in and enabled by default.

vdkeybus commented 4 years ago

Ok, in that case I will work with the exceptions in configuration.yaml.

This can be closed.

Thanks.

Op di 7 jul. 2020 om 22:12 schreef Koen Kanters notifications@github.com:

I'm wondering if this option shouldn't be turned off by default. Or even removed.

It will be removed in zigbee2mqtt 2.0.0, but as this is a breaking change it is kept in and enabled by default.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Koenkk/zigbee2mqtt/issues/3865#issuecomment-655102785, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSNXH5HLFZ5RYDAR6WZIHTR2N6SRANCNFSM4OSXGOLQ .