2022-08-26 14:47:29.351566+02:00
Seems to be right
Now it just triggered! But 7 minutes late?
From trace timeline:
“Triggered by the time at 26 augusti 2022 14:54:03”
What type of install are you running?
Running VirtualBox on a MacBook Pro
Home Assistant 2022.8.6
Supervisor 2022.08.3
Operating System 8.5
Frontend-version: 20220802.0 - latest
When looking in the history/logbook of that automation I see this:
(Translates to turned on/became unavailable)
Why would an automation become that?
Hasn’t triggered for one week now. Anyone know where to start looking? I’m clueless
Might be worth looking at resource usage because if a simple automation like that is not triggering, it sounds like the system is extremely busy.
As for why an automation becomes unavailable briefly, that’s normal, when you create any automation or edit one, and press save - all the automations are reloaded, so briefly they become unavailable until they are loaded back in.
The systems runs smooth. CPU and memory usage seems normal.
I tried adding the time date platform and when I create a card to show the time its very often of by 5-10 minutes. For example, now the time is 21:43 but the sensor.time says 21:36.
In the history the time sensor gets updated every 2-10 minutes (16:00, 16:01, 16:06, 16:12) Could it be that if I have an automation to trigger at 16:04 it doesn’t because the sensor never becomes 16:04?
Changed the automation to fire at 22:25 and it trigger at 22:55. What is going on? 
”Triggered by the time at 7 september 2022 22:55:33”
Yes, but that is resource usage from INSIDE the Virtual Machine. The actual host is a Macbook - and if that becomes bogged down, then the CPU cycles aren’t there to keep the VM running smoothly. But from inside the Virtual Machine it won’t show you any issues, because it has no idea about the system resources on the host itself.
If you go to the VM and type login on the command line, and then dmesg there is a good chance you will see messages complaining about the hard drive being busy or the CPU frequency or something similar, because when the VM wants to do something, but the host CPU isn’t able to allocate CPU time to the VM, then the VM will misbehave.
Is it possible to do a page break? When typing dmesg I get thousands of lines scrolling by so I can’t really see what it says
If you use dmesg | more you can get paging yes.
But if you use:
dmesg -T | egrep -i 'killed process'
You can limit to only lines that specifically tell you a process was killed by the system.
I only get:
/bin/ash: egrep: not found
Scrolled through it manually and can’t find anything about any killed processes. The system runs quite smooth. Shouldn’t it be very bogged down if it should miss a simple time trigger?
This seems to be the solution!:
Something to do with running virtual box and the “Paravirtualization Interface”.
Now the time sensor updates all the time and the automations seems to trigger every time. Will check some more before I mark this as solved
What setting was it on, out of interest?
I changed from default to none
The link you posted points to the wrong post in the other topic. The one mentioning VirtualBox’s Paravirtualization Interface is the next post.
For future reference, here’s Oracle’s documentation for the feature:
https://docs.oracle.com/en/virtualization/virtualbox/6.0/admin/gimproviders.html
Another useful post:
Interesting.
I don’t know what Default - equates to.
KVM is recommended for Linux guests.
On starting the VM it picks the Paravirtualization interface based on the guest OS it detects (so it’s an auto-configuration mode).
1 Like
This probably explains why I sometimes had problems logging into HA using Google authenticator