New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Xiaomi Aquara Water leak sensor (SJCGQ11LM) falsely reports trouble on deConz 2.22.2 (HA addon) #7061
Comments
@SwoopX can you check? |
Same here. 3 of 4 aquara water sensors started reporting water shortly after update to HA addon 6.20.0 yesterday (deCONZ 2.21.2->2.22.2) running with ConBee II If additional logs are necessary I can extract them |
We changed the DDF in #6919 to use the Xiaomi special attribute report next to the IAS Zone. This PR was merged in v2.22.1-beta. I’ve been running that version since it’s release. Haven’t upgraded to v2.22.2 yet, but I don’t think that version has additional changes for the water leak sensor. Pressing the button causes a special attribute report to be sent. Could you turn on logging (I think it’s APS level 2) so deCONZ lists the payload of incoming messages on a separate line with I think I saw some false positives just after installing v2.22.1-beta, but these were caused by an almost empty battery in the sensor. The value of Make sure to test the new battery before installing it to the sensor - it should deliver at least 3.0V. I’ve seen new button cell batteries that already lost too much power during storage. |
@bennydiamond same problem as me. but unfortunately I didn't manage to fill out the form, so it was closed to me. By the way, I first reported on Discord a month ago... EDIT: the special thing is that there is always about 50 minutes between announcements EDIT2: @ebaauw my battery has 2.92-2.90V in one sensor, but changed for new with 3.21 and will wait :) EDIT3: testing new battery and it's same. I also tested restarting the container and then pressing the buttons on the sensor and it also declared an alarm. Log:
|
The sensor reports Model Identifier (0x0005) and the special attribute (0xFF01) in one message, in response to pressing the button.
deCONZ is working as expected; the sensor does actually report leak detected through the XIaomi special attribute. Battery should be OK at 3.045V, though. Note the older firmware (my newest sensor has 0.0.0_0006, however this model doesn't seem to support firmware updates, not even through the Aqara M2 hub). Also note that the reported length of the special attribute value is wrong (34 bytes reported vs 33 bytes sent), but that's the case on my sensors as well. Did you try shorting the contacts for a while? Does the detection state change when you remove the shortcut? I vaguely remember messing with the sensor for a bit, after replacing its battery, but I haven't kept a log of what I did exactly. I might have done a factory reset and re-pair to deCONZ. If everything else fails, you could delete the
That's the interval of the regular special attribute reports. |
Look at my closed bug report: #7059 I just wrote here that I have several water leakage sensors. Two of them have a different FW version than the other three (I also added screenshots there). Now older sensors are 0.0.0_0004, but were 20170721. Newest are now 0.0.0_0006, before upgrade 20200310. As soon as I'm at the device, I'll try artificially shorting the sensor and send the log if there's any difference in it. I'm sorry, but I don't know what DDF is and how I can delete the given parameter from it, I'm just a user ;) EDIT: logs: All time:
When sensors are connected:
|
Please see https://forum.phoscon.de/t/aqara-leak-sensor-new-firmware-unavailable-entities-in-ha/3702/14. Summary: since updating to 2.22.2 (deCONZ addon for Home Assistant) and restarting the container, two entities are missing: Where's the old DDF? I'd like to use that. Oh wait: why has this been removed?!?
Guys I spent hours trying to bring things back and now I read "ah, well, it has been removed" - which is NOT listed as "breaking change" but as "ENHANCEMENT" ( |
Well, imho, it's not a breaking change as the PR doesn't break the two items from the screenshot above. You cannot break what non-existent on the device. |
So there's never been e. g. a low battery feature? What I see is:
So technically something broke. If it never was provided from the hardware side so why was it provided in the past... than the PR would be a correction which is good - but the communication is not sufficient, that's the point. |
I'm sorry, but again: you cannot break what's not there. Of course, the question is valid to ask why it was included in the first place but you probably will never get a (satisfying) answer: it was added by an inexperienced user just expecting the feature to be available, items were added as defaults to a certain device category automatically, features were assumed based on the relevant cluster and the device following zigbee standards, accidentally, etc... we're all humans in the end. Corrections/updates on devices happen all the time, similarly to other projects. They way of communication or level of detail, respectively, is certainly always debatable. In my view, this is also not given for other projects, but I might be wrong here. Not sure if mentioning more details in the release notes would be a silver bullet as they are not frequently checked in my experience. Anyway, we're drifting away from the topic of this issue, so we should keep it focussed. |
The problem is not the tampered and low battery entities, is the most important one which is leak detected! :) |
I've got same problem. |
I only have sensors with version 0.0.0_0004 firmware and they exhibit the issue. |
I have sensor with version 0.0.0_0003 and that problem. |
I have 4 Xiaomi motion sensors, they all give me now false detections, one almost every hour, the other one like every 6 hours... Seems all happening after the last update (using ha addon ) |
@pergolafabio please do not highjack this issue, however, feel free to create new one. Thanks! |
I thought it's maybe related to Xiaomi detection in general? |
Its not. We recommend opening a new one. |
Got a (presumably false) notification from my 0.0.0_0004 sensor today. Looking at the legacy code, we used to ignore the Xiaomi special attribute for this sensor. So it shouldn't have been added it to the DDF - my bad. Line 888 in 2f021e0
I'll prep a PR. In the meantime, as workaround, you can remove the |
...until restart docker. Now is it back. How can I edit DDF please? Right click, EDIT: OK, edited directly in shell.
and deleted |
I only have sensors with version 0.0.0_0004 firmware and the same issue since upgrade.... |
deConz 2.22.2 (HA addon) I don't have this location: |
Hi, I tried using your fix with the 2.23.0 beta branch, and my one leak sensor with firmware 0.0.0_0004 (all other 9 I have have firmware 6) is still reporting a false leak. Is that expected after the fix? Is there a way to make the false leak go away without physical interaction of the sensor (i.e. shorting the leads, or re-pair it)? |
Just to mention the Aqara leak sensor does not have a firmware version issue. The issue with missing entities for leak and door sensors is caused by the bad ddf in the Lidl watering system see https://forum.phoscon.de/t/aqara-door-sensors-entities-not-available-in-ha/3949/16 |
Hi and thanks for mentioning. However, we should highlight that a missing property of a different device does not affect the Xiaomi leak sensor and therefore, deconz REST API in any way. If a REST API clients gets confused based on that, it's unfortunate. |
Any news on this topic? I did a complete new system installation, but still having the issue with my water leak sensors (both with firmare 0004) reporting a false leak. Initially I thought they are working now, but after a reboot of the system, since more than 1 hour false leak reported. |
I have 10 leak sensors and none of them received any fake alert. |
Describe the bug
I'm running deConz HA addon.
I updated addon to version 6.20.0 (which includes deConz 2.22.2). One of my water leak sensor SJCGQ11LM started reporting trouble right away.
I tried to delete and re-pair the device, device will behave normally for a couple of seconds and then will fall back to reporting trouble.
I tried replacing the battery in the device, same issue.
I downgraded deConz HA addon to previous version (6.19.0). The problem went away. SJCGQ11LM sensor behaves as intended.
Only 1 of my 5 * SJCGQ11LM sensor exhibited the problem but I suspect this is due to the fact the others might have been sleeping and thus did not report to deConz while I was witnessing the issue.
Steps to reproduce the behavior
Pair "SJCGQ11LM" with deConz HA addon 6.20.0
Once paired, press the button on SJCGQ11LM sensor to trigger sending its attributes
Device will report trouble/wet and will stay in that state indefinitely
Expected behavior
If no water is physically touching the sensor, the sensor should not report trouble/wet.
Environment
Additional context
I reverted to HA Addon 6.19.0 and did not save any logs. If necessary, I could update again to the faulty version and generate logs.
I highly suspect the issue originates from this change: #6919
The text was updated successfully, but these errors were encountered: