2022.7: A stunning performance

Has anyone informed McDonalds or The Old Spaghetti Factory??

3 Likes

I saw your issues with REST sensors and matched with my long long loading time for my REST sensors gave me the idea to remove them. The result was I was no longer having memory leaks. Although the memory is higher overall, it is sustainable (60% of Rasp Pi 4 4GB) rather than leaking out and then crashing every 12-24hrs.

Yeah, I figured out that the problem was ffmpeg older version. I run home assistant on RPi 3. Hopefully bullseye will work on it. Thank you, anyway. Iā€™ll give it a try.

Iā€™m on a RPi 3 as well and it went fine for me. Took a while, obviously, but it worked. Make sure to do the DHCP part in that guide otherwise you wonā€™t get an IP address from DHCP when you reboot your pi. Best of luck.

seems hardly an issue, why would you think this would not cause an issue? how would a system be able to distinguish between 2 files with an identical nameā€¦

tbh, a base rule in computer programming would be to give uniquely identifiable names to anything really. Packages not excluded.

the split files are re-united by the backend yaml parserā€¦ and even the file editor would want you if youā€™d safe a file with an already existing name?

Really?
C/Java - I may have lots of files with same names, if they are in different folders. Probably the ā€œunique file namesā€ rule exists for some languages, I donā€™t know.

Why the system should use a packageā€™s filename?

I get it if you refer to letā€™s say the manifest.json files etc in Custom components.

maybe youā€™re right if this would be comparable. yet, the packages are mainly for us humans to be easily organized, in the backend they all are made back to 1.

One could argue the nested folder in which the identically named package file is saved would ā€˜belongā€™ to that file(name). Allowing for many identical file names, as long as they are in different subfolders.

not sure, guess I never ran into the issue, because I want to be able to see whatā€™s in the file myself, by looking at the file name. So not a single identical name in my Packages (sub)folder(s).

Lets see what response youā€™ll get in the issue.

Not exactly, I meant lots of ā€œ.hā€ & ā€œ.cppā€ files in C/C++ projects, for instance.
They may have same names. Surely filenames must be unique within a folder, this is a restriction of a filesystem.

As for HA - probably each package (i.e. yaml-file) may be processed not by itā€™s filename but by itā€™s unique ID, if it is important.

A small use-case: in the ā€œgarden waterpipeā€ folder you may want to have a ā€œtest.yamlā€ for testing something. Why cannot you have another ā€œtest.yamlā€ in a ā€œkodiā€ folder?

yeah. thing is, I like to move my test files in and out of the real ā€˜Packageā€™ folder, and move it to a non included Put aside folder and do a reload.

Returning the files after that would make it rather hard to do without anything furtherā€¦ thats why I would name these files package_garden_waterpipe_test.yaml and package_kodi_test.yaml which would allow me to move them back into place without checking the contents

I am using a different approach.
Assume there is a folder with subfolders & files (let it be only yaml-files for simplicity).
If I do not need some folder - I just do this:

  • add a ā€œ.disā€ (ā€œdisabledā€) suffix to the foldersā€™ names;
  • add a ā€œ.disā€ suffix to the yaml-filesā€™ names.

The 1st step is just to specify a different name for a folder, does not affect on excluding a yaml-file from a package.
In the 2nd step - the renamed yaml-file is not considered as yaml and then is excluded from a package.
When using a ā€œsearchā€ in Visual Code Studio, I always see a location of some code whether it is in a ā€œyaml-ā€ or ā€œdis-ā€ file.
And I have all old or temporarily disabled solutions in a one corresponding place.

Hmm, besides me not completely understanding why you are doing it the way you are doing it, I guess this would for me be ā€˜why do things in a simple manner if you can do things difficultā€™

Simply moving a file out of the packages folder is simplest as can be.

But hey thatā€™s just me.

Restore has saved my system more than once. I am no longer afraid of screwing around and trying things. That said, Iā€™ve never done a partial restore.

1 Like

Since you donā€™t understand, itā€™s ok.
Some other people prefer things more structured.

Change the files from .yaml to .yml. As noted from the docs:

We offer four advanced options to include whole directories at once. Please note that your files must have the .yaml file extension; .yml is not supported.

This will allow you to !include files with .yml extensions from within the .yaml files; without those .yml files being imported by the following commands themselves.

That way they get skipped by the directives but also you can still get syntax highlighting when you open them in an editor. And can include them manually if you want. Thereā€™s no equivalent to your folder one though.

On your issue:

  1. itā€™s not a bug, that should be a feature request. Directives are working as designed in a way they have been for a long time. Youā€™re requesting a new feature - support for repeated names in different subfolders.
  2. you should exclude !include_dir_named as that doesnā€™t make sense. That directive uses the file name as the key in the object it creates. If that key became the path to the file then it would be like folder/folder/automation and that wonā€™t really work
  3. itā€™s a huge breaking change to advanced configs. I canā€™t imagine that being implemented for that reason tbh. Think youā€™d be better off requesting a new directive that works the way you want so existing configs donā€™t break.

This is a good workaround. Some day I managed to setup my Visual Studio Code to treat these ā€œ*.disā€ files as yaml - just to have this highlighting. But now I live again w/o that highlighting (switched it off) - and I easily see ā€œthis file is disabledā€.
The only thing is - looking at the fileā€™s list in VCD, it will be not easy to differ ā€œactiveā€ files from ā€œdisabledā€ files, they are of same color:
ŠøŠ·Š¾Š±Ń€Š°Š¶ŠµŠ½ŠøŠµ

ŠøŠ·Š¾Š±Ń€Š°Š¶ŠµŠ½ŠøŠµ

Probably there is some way to set different colors for yaml & yml files, I will check.

Itā€™s ok then.
Users like me then just will keep setting unique filenames.
In my scenario: ā€œgargen_waterpipe_test.yamlā€, ā€œkodi_test.yamlā€.
Yes, this is not convenient, but itā€™s ok.

Also, repeating filenames should be logged as an error - which does not happen now.

Since files may be located in some subfolders - then probably these ā€œkeysā€ could include these relative paths, so they become unique. Ok, if you say that it is not possible - I just have to believeā€¦

Everything is possible but it has to be clearly defined. Right now the directive collects all the files and uses their name as the key of the object it makes. Youā€™re suggesting the same file in different subfolders still works so I guess I assumed that means the path becomes the unique identifier in the object. It could be different though.

Thats why Iā€™d suggest just writing up an FR spelling out clearly what you want to happen.

1 Like

Right, I will do it. Here is a FR.

Thank you for explanations.

Then this issue is about ā€œshow an error/warning in case of repeating filenamesā€.

ā€œRESTARTā€ and "CHECK CONFIGURATIONā€ spin forever, itā€™s not working. Iā€™m rolling back :ā€™(

I removed miflora, and now itā€™s ok. I switch to Passive BLE Monitor integration

Stunning performance?
On what machine?
When I look at ā€˜system statusā€™ I see,ā€¦
Home assistant supervisor 10.70 s
Zone 12.92 s
Mqtt 13.25 s ( for only three devices)
Input_select 13.54 s
Input_button 13.55 s
Raspberry pi 13.59 s
Input_datetime 13.73 s
Input_boolean 13.74 s
Input_number 13.80 s
Nest protect 14.77 s
Hacs 15.80 s
Version 20.26 s
Switch_as_x 24.93 s

On a rpi 4, 8gb ram

Is this stunning? What numbers do you have?
Maybe time to buy a nuc.

24 second startup time is very low. You miss understand those numbers. They arenā€™t additive. They are how long it took for that to get set up from startup.

going off the numbers you postedā€¦:

Home assistant supervisor took 10.70 secondsā€¦
Zone took 2.22 secondsā€¦
input_select took 0.62 secondsā€¦
input_button took 0.01 secondsā€¦
Raspberry pi took 0.04 secondsā€¦
Input_datetime took 0.14 secondsā€¦
Input_boolean took 0.01 secondsā€¦
Input_number took 0.06 secondsā€¦
Nest protect took 0.97 secondsā€¦
Hacs took 1.03 secondsā€¦
Version took 4.46 secondsā€¦
switch took 4.67 secondsā€¦

2 Likes