This is surely common knowledge for some of you but I thought I'd share a story, as this just made me cringe
I never understood the point in the option to "run actions in parallel" I thought if I had a list of actions to complete, HA makes it through them almost instantly, and with the varying latency of each action they wouldn't complete at the same time anyways.
Then I tested my smoke alarm notification that I have had running for over a year.
It went-
If: list of smoke alarms detects smoke Then: Turn on the lamp next to my bed, then Send a notification to my phone.
I had made an error when setting up the lamp entity. (I made it full brightness on both scales, can only use one) this stopped the automation before the notification went out to my phone.
If it's important that the automation makes it to the end, run in parallel!
Talk about a false sense of security
I gotta be honest, I'm not sure I'd be willing to trust something I set up myself with general-purpose software to handle something as important as a smoke alarm alert.
That's the sort of thing that gets hardware dedicated to the task and doesn't rely on me configuring everything correctly and Linux not crashing because some other unrelated process had issues.
save