2021.4: For our advanced users ❤️

https://github.com/esphome/esphome/tree/c00c2d43bb7a5c1a0f5f6837e1c084327ff21c72 doesn’t seem to exist…

But this one does :stuck_out_tongue:

(which is not a exactly a link to a commit, but an archive tree link from april 7th :stuck_out_tongue: )

Since update from 2021.4 to 2021.5, the following message continues to appear. How to resolve?

1 Like

found my issue, my MariaDB wasn’t fully running
i was able to connect by phpmysql, but it seems my data was going nowhere
after force a restart of the mariaDB on my synology, i’m seeing my states back.

Since release 2021.4.4 (I assume with the upgrade of maxcube-api to 0.4.2) HA maxcube integration blocks all further connections with other apps and also the Windows app for the configuration to the cube. If you restart HA completely and catch the right moment, you can log in with other apps. After that no more.

well not issue is still happening just stored a few minutes after DB Restart

I have the same but reporting the core-2021.3.4 as the latest version even I’m running core-2021.4.6 version.


1 Like

I am using Homeassistant now for a year. But to be honest: What are you guys doing? Every second or third so called “update” breaks the installation and one has to start from scratch, or from a backup (if you have one).
I’m using a RasPi AND a virtual instance to check every update, before putting it “in production”. But even that reveals silly bugs and renders the Pi-installtion broken.
Is anyone TESTING that stuff, before releasing. Or is it just me with that experience?

After this update one can’t even use the supervisor to start/restart or run a restore. It’s really frustrating.

interesting statement, never had to start from scratch other then only once a fried SD card. What does break on your setup according to you requiring to start over?
‘This update’ especiialy has a heavy DB upgrade, did you let that finish?

No real issues here either…
I only had to re-install my back-up once, but that was mainly my own fault. Took me only 15 minutes though…
Running tasmota, mqtt, esphome, zigbee and a custom integration of my own

It is true every now and then i have to adept my configuration but i am aware of it as i always read the release notes, especially the breaking section

I was not able to use the supervisor(GUI) this time. That happened before (without update), too. So, there could be something else be wrong. also.

All sorts of things happened with updates to me in the last year. Once there was this situation, that I could not use the backup after an unsuccessful update in the “onboarding process” while reinstalling, because something was changed here and the compatibility was not given anymore. Don’t remember the details, sorry.
But after that I startet testing it in a virtual machine, before running it on the Pi.

Yes, I did let the DB upgrade finish.

Anyways, since I have the Pi, a virtual machine AND a daily backup I am up and running in less then 15min, after situations like that. And a snapshot in the VM is really nice, too. No problem here, it’s just a little frustrating.

Besides that: When it’s working I am happy with all the functionality given and the the possibilities in integrating all sorts of things.
AVMFritz, deConz, HomeKit, MercedesMe, MGTT, Panasonic Comfort Cloud, Shelly, XiaomiMiio, 433MHz switches, Synology and some own integrations. Maybe it’s just too much stuff.

So, how do you have it installed? All I see is ‘virtual machine’. But you haven’t listed your installation method. HassOS, Supervised, Container, or Core?

FWIW, I’ve been using Home Assistant for almost three years. I’ve performed dozens of updates and never encountered one that 'breaks the installation and one has to start from scratch, or from a backup".

Yes, but not like commercial software is tested. Whoever volunteers to test Home Assistant’s beta version (a fraction of all users) does and then, when the reported bugs are fixed (over a 1-2 week period), it is released. A much larger number of people run the major release version and its remaining bugs are detected and corrected (in patch releases). All of this happens monthly.

So if you feel a bit like a beta-tester when you use a release version, you kind of are because the paucity of volunteer beta-testers means some bugs will only be discovered by the broader audience using the release version.


Well, never fully understood what is what.

What I did was - wrote the image (5.13/hassos_rpi4-64-5.xx.img.xz) on a SD and started the “onboarding” for the Pi 4. → Pi is running
Used the hassos_ova-5.xx.ova for the virtual system and recovered a Pi backup in that onboarding process. → testsystem is running.
From here I was using the backups and snapshots I have written.
Guess I am using HassOS?

Yes, it looks like you’re running HassOS on your PI, and HassOS on your VM. I’m assuming your VM is on a separate machine from your PI…

Yes, that is a separate hardware, emulating everything the Pi does.

Both of those routes are pretty well tested, even with updating. I’m not sure how it can break every version for you.

Well, thank you guys for your answers.
Got it running now … don’t really know what (I) was (doing) wrong.
Should not have released my frustration here in the forum. I really was frustrated, this time.
Sorry for that.

1 Like

Just you .

1 Like

No problem, can’t hurt to check others experiences.
Just a tip, try to read the release notes (especially the ‘ breakings’) to see if the update would affect you, and if it does, try to prepare on what changes should be made (these changes often can be done before the actual update :wink:)

I had my System on a VM running On Synology, my DB is MariaDB 10 also running on Synology, i also store my data on influxDB when 2021.4 came both my DB were around 20gb each, past around 7 hours on migration so i decide to start a new SQL DB because i had my historic data on influx, on update 2021.4.3 my log grow until 11gb, had to update all my Shellys to stop that to increase.

On 2014.4.4 i notice that my states just track for a while and then flat line on the state graphic but the state value still changed and been storage on influxdb, but is not tracked on the MariaDB

I’m also not having like connection issues or something else to track this issue, sadly the closest snapshot did i had of my VM is on 2014.4.4

i think i’m going to start all over again