Add a home assistant start trigger that checks against the time. That’ll solve that problem.
Edit: Marius beat me to it
Edit2: I think your original template would have worked. With just a restart time. However it wouldn’t have worked if you restarted between 22 and 23. So if you just changed the time check to look at 22:59.9 you’d be good to go
Nevermind I’m taking crazy pills
Hi.
Just an update on the TOD problem. It is as I expected. The problem occurs when you have a TOD type binary sensor with these conditions.
The result is that the sensor goes off after a HA restart
@petros briliant template sensor works great. It turns off for less than a minute and then back on again. Used as a condition in an automation that is not a real problem. The automation is not working while you restart either. It is all back within a minute.
@petros. Challenge.
This also fails
- platform: tod
name: Darkness
after: sunset
before: sunrise
You have an idea for a template work around for this? 
Anyone seeing issues with streaming camera videos after ~2021.5 update? I’ve got 2 foscam’s that as you watch them lag/reload periodically and you can see the time stamp on the video roll back about 10 seconds. If you let the stream play continuously, it basically never catches up and keeps going back in time to a point where you started about 10s behind live and get up to 30min + behind.
After updating to 2021.5.3 the automation below stopped with error.
- id: lights off
alias: 'lights off - 23:00u'
trigger:
platform: time
at: "23:00:00"
action:
- delay:
minutes: '{{ range(01, 30)|random|int }}'
- service: light.turn_off
entity_id: light.gang
- service: light.turn_off
entity_id: light.woonkamer
Logger: homeassistant.config
Source: config.py:443
First occurred: May 12, 2021, 11:00:12 PM (2 occurrences)
Last logged: 10:13:07 AM
Invalid config for [automation]: expected float for dictionary value @ data[‘action’][0][‘delay’][‘minutes’]. Got None. (See /config/configuration.yaml, line 37).
After removing the delay the automation is ok again.
- delay:
minutes: '{{ range(01, 30)|random|int }}'
Edit: solved after changing 01 → 1
- delay:
minutes: '{{ range(1, 30)|random|int }}'
Reply to my own challenge earlier.
So situation is that this binary_sensor fails (it is off after a HA restart right after midnight)
- platform: tod
name: Darkness
after: sunset
before: sunrise
Unless the sun integration is also been affected by the same bug as TOD (Time Of Day) this may work
You create a template binary sensor like this
- name: "Darkness"
unique_id: darkness_template_sensor
state: >
{{ is_state('sun.sun', 'below_horizon') }}
I need to verify tonight if this gets the correct value a few minutes past midnight after a HA restart. If so this is a simple work around.
Edit. I verified that the Sun based sensor works fine and not affected by the TOD bug so my Darkness sensor above works
Today at 2:54 am the cpu temperature fell to 35 degrees. Can‘t explain it because the only service that was running at that time was deconz.
I am having a problem after upgrading to 2015.5.3
Home assistant keeps on restarting.
I have the Supervised installation on Banana Pi Pro running Armbian
Armbian 21.02.3 Bionic with Linux 5.10.21-sunxi
The docker image installed is as in the image below
Hmm… for me after performing at least 15 of HA restarts I need to summarize:
Where a HA restart on HA Core 2021.4.6 took only up to 2 minutes for 4 out of 5 reboots,
now on HA Core 2021.5.3 one single reboot takes at least 6 minutes for 5 out of 5 reboots (100 % of time).
So stability (maybe) +1, but performance -3 ?
I am also having somewhat similar issue. Upgraded from 2021.5.2 to 2021.5.3 and over the night CPU usage shot to 100% on Proxmox VM. Server then crash. Revert to 2021.5.2 and all back to normal except CPU is still slightly more than 50%. For previous 2021.4.x versions, CPU will idle around 10%. Not sure what is causing that. 
Update: After 12hrs of reverting to 2021.5.2, CPU is at 10%.
How do I downgrade.
Revert the snapshot…
Or pick the version you want from the command line, e.g.:
ha core update --version 2021.4.6
With Proxmox, it is really easy, i just restore a previous version snapshot. 2 secs job! 
You probably have to use the other methods suggested by the other posters above.
How much time is needed for the database migration process? It’s more than 24hrs i see the database temporary files (-shm and -wal files) updating every minute…
Is it normal?
It depends on your DB size and HA server specs.
My NUC only took about 18 mins to do a 3.1Gb database.
2021.5.4 looks to have completely broken the remaining bits of FritzBox_Tools (the custom component, not the new integration).
So currently no switches for profiles or wifi, which haven’t yet been moved into the new integration
Rolled back to 2021.5.3 and all working again. Hopefully it can either be fixed or rolled up into the new integration…
It’s best to report that to the custom integration. Reporting it here doesn’t help because the custom integration creator may not see it.
Already did that as well.
Looks like something went weird with the update from 2021.5.3 to .4, as rolling back and doing it again seems to have cured everything.
anyone else noticing frequent network issues lately, I believe since 2021.5.2?
these errors are frequently showing, and all point to images in my www folder:
and have not changed since a couple of years 
I’ve enlarged some of them (not these btw) to make the pinchable, but hardly believe size matters in here?
Do believe it is mostly Safari, my Chrome browser and the App show fine, where Safari blanks on these images (used on custom card Button).
