Does not work on HA Core 2022.05. #129
Comments
|
Same here. Totally blown up |
|
|
|
Logger: homeassistant.helpers.frame Detected integration that accesses the database without the database executor; Use homeassistant.components.recorder.get_instance(hass).async_add_executor_job() for faster database operations. Please report issue to the custom component author for average using this method at custom_components/average/sensor.py, line 477: history_list = history.state_changes_during_period( |
|
Seems to me that developer stopped maintaining the repo - he didn't respond to my previous issue regarding database executor also. So I'm deleting the addon :(. |
|
any alternatives? |
Not that I know of... but as I already had to change a bunch of automations and templates in HA as average sensors stopped working (I had a good dozen of them) I'll now simply remain this way. Averages were useful, but not worth the hassle regarding latest issues. |
|
|
That's very sad as this integration allowed to smooth out values or averaging multiple sensors. Need a replacement solution for this :'( |
|
Need a replacement for this too! |
|
I am in no means a python expert but looking at the line where it is failing: https://github.com/Limych/ha-average/blob/dev/custom_components/average/sensor.py#L491 I found out that that function |
For averaging multiple sensors you can use min_max integration. For an average you can use statistics->mean but I already have problems with large averages (+24h) since 2022.4 and I use this platform. There are also a long-term statistics that I really don't know how to use |
Yes, this is my first stage of averaging. That way I can have multiple sensors in large rooms (e.g.: garage)
Yes, this is what I switched to for my second stage of averaging in order to eliminate wide variations (e.g.: temp sensors with low resolution). The problem is that it's not possible to average let's say the last 15 seconds to smooth out the value. If the sensors don't change value within that timeframe, then sensor goes unknown and all my dependent sensors, binary sensors, automations, dashboard gauge, climates, everything goes wild.
Those seems to be for dashboard history-graph as what I understand and won't be helpful in my case |
|
same problem here! |
|
same here. |
|
check out this one guys... https://github.com/jeroenterheerdt/HADailySensor I think this fixes my requirement for average and min/max Actually not really as it won't have attributes with min/max. Doh! |
I just started using this yesterday and so far it meets my needs. You can do min/max by creating separate sensors for those |
Yes I know that but I use the flex state card which I can select attributes of an entity to display in a table but I can't specify multiple entities. |
Probably not ideal but could you create a template sensor with all the attributes you need? Might work until you can find a better alternative or this gets fixed |
|
I don't think you can create a sensor from a template with attributes. I'm all ears if you have an example |
Yes you can, I use tons of sensors with attributes... on my phone right now, but the template sensor documentation explains it with examples: https://www.home-assistant.io/integrations/template/ |
|
Yup, you definitely can. I'm on my phone, so the formatting might not be perfect, but you just add a map of attributes to the sensor you're creating. Something like this: Since each attribute is a template, you can populate them with basically any value you want. |
|
I i just had a quick look at the profile of @Limych . Says there he is in Moscow, Russia.... means... he might be locked out of GitHub until the situation in the Ukraine has been resolved... :-/ |
Me too from Moscow, and no personal sanctions on GitHub. I can use it right before of all this madness. |
|
Quick tip: Besides the min/max and statistics integrations, there is also "filter" the combination of these 3 gave me everything I need to replace this dysfunctional integration. |
How did you replace "process_undef_as" functionality? |
I don't. for me there is no need.
min/max - mean makes an average/mean of multiple entities. so multiple sensors into min/max. the result put into statistics and/or filter. come to think of it, you can template for the process_undef_as with: {% if is_state('sensor.source_abcd', 'unwantwed value 1 like unknown') %} 'wanted value' |
The built-in functionality updates the calculated values only when the input sensor changes, so when the input sensor reports the same value for a long time, eg. the average value will remain an old/invalid value and never gets equal with the input sensor. This makes the built-in versions nearly useless (for me). OK, I'm tested this approx. half a year ago, but I don't think this is changed. |
|
I added the get_states back into the init.py and recorder.py files. Everything working now until next time I'll update.. Now I have some time figure out what to do |
So you forked it and fixed it? anything you want to share? |
|
Yes, no problem! I am using docker, but locate home assistant core folders and go into components/recorder. In init.py add this at line number 71: in history.py add this at line numer 473: and this at the bottom: Reboot and you should be good until next time you update |
|
Is there a way to change the custom component's file instead of the HA file? |
|
Wow this is a huge issue for me too if it doesn't work. I hope someone wants to fork it and continue the work. I'll be happy to contribute if I can. |
I'm done it: https://github.com/lmagyar/ha-average/commit/acd751587609951a1baf41e6a2b162b0fc0cf3e1 And don't forget to restart HA core!!! During startup the previously sporadic errors like some sensor is Undefined and cannot be converted to float now changed to Disclaimer: I'm just an average HA tinkerer, this is just a workaround, but if it is working for others I will create a PR and somehow "save" this project in HACS. |
It worked for me! Thanks @lmagyar |
Works for me too
No, you're great, Thank you @lmagyar... |
|
Worked with the fix of @lmagyar Let's see if the developer can update their code... |
|
Thank you. This made it work again! |
|
With your fix I get an error. "The average platform for the sensor integration does not support platform setup. Please remove it from your config." Why is that? |
I have absolutely no clue. 😲 I'm quite a beginner in HA hacking. 🥺 Though it seems to me some yaml/config error. |
|
I didn't change any config since 2022.5, don't think my yaml is the culprit |
|
For example:
Worked for ages... Now everything is broken 😭 |
|
Hmmm, I guess something is broken with your sensor.py, this should have been downloaded: https://raw.githubusercontent.com/lmagyar/ha-average/acd751587609951a1baf41e6a2b162b0fc0cf3e1/custom_components/average/sensor.py |
|
Yep, it was broken... Downloaded and copied again, now it works again! 🤩 |
|
It's work!! Great!! Thanks a lot!! |
|
Just info: If somebody noticed that due to this "transitory" hiccup with this component, the averaged values lost their Note: If you delete data from recorder database, this is your responsibility, only execute that SQL statement if you understand, that all those values are gone forever. And who knows what are the unintended consequences. |
|
There is no problem in execution, but the following log is displayed in the core. Could this problem be solved? Logger: homeassistant.helpers.frame Detected integration that accesses the database without the database executor; Use homeassistant.components.recorder.get_instance(hass).async_add_executor_job() for faster database operations. Please report issue to the custom component author for average using this method at custom_components/average/sensor.py, line 477: history_list = history.state_changes_during_period( |
|
I've been hit by this as well and not willing to live with a hack in my productive system. I am rather proficient in Python, but not in the whereabouts of custom HA integrations – the boilerplate scared me off. So I implemented a time-weighted rolling average function node with right Riemann sums in Node-RED. Just in case someone else also wants an easy-to-understand and clean solution and uses Node-RED along with HA. It is straightforward to create an averaged HA sensor with an [{"id":"c3195a8d84f95c56","type":"function","z":"f2681015255691ca","name":"time-weighted average (over 5s)","func":"/*\ntime-weighted average of payload values\nover the specified time range (in ms)\n*/\n\nlet now = Date.now();\nconst range = 5 * 1000;\n\nlet t = context.get('t') || now - range;\nlet T = context.get('T');\nlet F = context.get('F');\nlet buffer = context.get('buffer');\n\nlet dt = now - t;\nlet increment = {dt: dt, df: Number(msg.payload) * dt};\n\nT += increment.dt;\nF += increment.df;\nbuffer.push(increment);\n\nwhile ((T - buffer[0].dt) >= range) {\n T -= buffer[0].dt;\n F -= buffer[0].df;\n buffer.shift();\n}\n\ncontext.set('t', now);\ncontext.set('T', T);\ncontext.set('F', F);\ncontext.set('buffer', buffer);\n\nmsg.payload = F / T;\n\nreturn msg;","outputs":1,"noerr":0,"initialize":"context.set('t', undefined);\ncontext.set('T', 0);\ncontext.set('F', 0);\ncontext.set('buffer', []);","finalize":"","libs":[],"x":490,"y":380,"wires":[["2bd98b77a5b589d1"]]},{"id":"e846aa77ca8fc389","type":"inject","z":"f2681015255691ca","name":"","props":[{"p":"payload"}],"repeat":"","crontab":"","once":false,"onceDelay":0.1,"topic":"","payload":"0","payloadType":"num","x":270,"y":360,"wires":[["c3195a8d84f95c56"]]},{"id":"c6d55e6e3484ea4e","type":"inject","z":"f2681015255691ca","name":"","props":[{"p":"payload"}],"repeat":"","crontab":"","once":false,"onceDelay":0.1,"topic":"","payload":"1","payloadType":"num","x":270,"y":400,"wires":[["c3195a8d84f95c56"]]},{"id":"2bd98b77a5b589d1","type":"debug","z":"f2681015255691ca","name":"","active":true,"tosidebar":true,"console":false,"tostatus":false,"complete":"payload","targetType":"msg","statusVal":"","statusType":"auto","x":730,"y":380,"wires":[]}]Start tab: context.set('t', undefined);
context.set('T', 0);
context.set('F', 0);
context.set('buffer', []);Function tab: /*
time-weighted average of payload values
over the specified time range (in ms)
*/
let now = Date.now();
const range = 5 * 1000;
let t = context.get('t') || now - range;
let T = context.get('T');
let F = context.get('F');
let buffer = context.get('buffer');
let dt = now - t;
let increment = {dt: dt, df: Number(msg.payload) * dt};
T += increment.dt;
F += increment.df;
buffer.push(increment);
while ((T - buffer[0].dt) >= range) {
T -= buffer[0].dt;
F -= buffer[0].df;
buffer.shift();
}
context.set('t', now);
context.set('T', T);
context.set('F', F);
context.set('buffer', buffer);
msg.payload = F / T;
return msg;I chose the right Riemann sum because it behaves nicely with initial values (I assume the initial value has been there for the entire time range) and last values (if the are no frequent updates, the output jumps right to the last value). |
It works! Thank you very much! Contact to developer to do as an official change |
This comment is a duplicate of #127. FYI: This is unrelated to this fix. And I was using this integration with debug level logging yesterday and haven't seen this warning, strange. |
I've contacted him even through email. There is already a ticket to remove this integration from HACS. See: hacs/integration#2654 |
|
Yes, I saw that issue. Sadly it's nearly greek to me also, + I can't reproduce the warning, + I have near zero knowledge about HA internals, + I even don't have a proper development environment. :( |
I'm reluctant to reimplement this integration if it's going to be dead next month anyway |
and others
Environment
Describe the bug
The component does not work on HA Core 2022.05.
Configuration.yaml
The text was updated successfully, but these errors were encountered: