Like I said before ,ssh addon doesnt give you access to the HassOs … You need to access the host on port 22222 , not 22 , install the other addon, or access it with the console directly if you attach a monitor
Thanks for pointing towards that port 22222. I am indeed running virtual, an will check out that addon to get into the host system on HassOS.
Last weekend I succeeded uploading an adjusted timesyncd.conf using the USB method. The problem I had earlier lied in the fact that I boot and run HassOS on an SSD. As soon as another USB device is plugged in, it won’t boot from the SSD anymore, but probably attempted to boot from the inserted USB and hanged.
Solution for me was to first initiate boot by powering on, and after a few seconds and blinking lights on the Raspberry Pi, insert the USB.
I have an Add-on called chrony which is in Configuration and under Add-ons, Backups & Supervisor. Clicking on the chrony Add-on go to the Configuration tab. You’ll be able to add ntp servers in that list.
Now to see if the ntp is working, I added 2 sensors to the Lovelace UI. They are:-
sensor.last_clock_synchronization which tells you when your machine last did a ntp synch and secondly sensor.date_time which shows the current date time on your system. I noticed on my Raspberry Pi it does an ntp action updating the time +/- every 10 min. When you reboot your device, the 1st time it does this, it thinks the time was set in 1970…which was 50 odd years back. The second time it does an NTP action, you’ll see it says it was done +/- 10 min back…
Immensely helpful thank you. I’m in a situation where only the organizations NTP server will work. Finally have my ha clock sync’d properly - boot time is drastically reduced as well.
How did you add those sensors? I can’t get my rpi to sync time, even tho the timesyncd.conf has been set correctly, and the servers are pingable from the rpi. Chrony tells me my host time is off by a few minutes, but still no sync.
Hi I was running some shelly devices OK for a while and decided to firewall them from the internet and only allow connection to my HA instance. All was working well until they lost time synchronisation and the local weekly schedules were not triggered anymore.
I installed the Chrony addon and changed my sntp configuration in each Shelly device to point to the HA server for time sync. It initially did not work, and I subsequently found out that this was because the “time zone and geolocation” part of the Shelly settings was on “automatic detection” and it failed (despite having pointed to the local NTP server).
I deactivated the “automatically detect…” box, had the timezone set manually, but left the daylight saving time on “auto”. I also set the longitude and latitude manually.
After reboot the time synced again and it is the right time (winter time). I don’t know if it will work in the longer run or if some cache will expire at some point, but I hope that this will work permanently if it could sync after reboot.
I ran a fresh install of 2024.7.4 on an Odroid N2+ and have spent hours scratching my head over the wildly wrong time it showed. I’m posting in the hope it helps someone else.
With a fresh flashed eMMC, it booted but would not run HA. After hours of searching and head scratching, I figured that the date was way off, causing cert checks to fail, thus no install etc. I set date manually which allowed me to boot, install and carry on. I was a bit puzzled at this, because I use DHCP for NTP and knew it was offering the local server which I use for all my gadgets, but the manual date set hack got me by.
A couple weeks later I picked up where I left off (migrating from RPi-4 - another story), and realised that the clock was off. Not a TZ issue, it was off by 33 minutes, and the Odroid has RTC (yes, battery inserted), so I concluded that it was drifting since I last set it (not very accurate) and not NTP synced after all.
That led me to find timedatectl (ssh as root on p22222), which refused to work ( “Failed to query server: Failed to read RTC…” ) until I ran ‘hwclock -w’ - I saw the hint elsewhere, no idea why this was needed. Then timedatectl worked and showed NTP not synced, and nothing I could try over the next 3 hours would get it to sync. Until I loaded Chrony (as above in this thread), which others indicated had solved an issue - and voila - instantly my NTP sync showed ‘yes’.
So thank you to the earlier posts in this thread - I’d still be going at it without the pointers. But now the question is why?! Chrony has a “Set system clock” option which is doubtless what saved me, but why do I need this; I don’t want another NTP server - why does the system not run synced to NTP, not boot without NTP or even a network, and not need another server add-on (which I don’t want) to make the system clock work?
If I can spend more time tinkering and figure out how, I’ll revise my post. In the meantime, it’s here because HA on Odroid N2+ with the standard image had NTP issues the moment I loaded it - others MUST be having the same issue…