Automation in restart mode - All runs cancelled about core HOT 11 CLOSED

I can also confirm the issue with an input boolean and this automation:

alias: Test Restart Automation
description: ""
trigger:
  - platform: state
    entity_id:
      - input_boolean.test_input_boolean
    to: "on"
  - platform: state
    entity_id:
      - input_boolean.test_input_boolean
    to: "on"
  - platform: state
    entity_id:
      - input_boolean.test_input_boolean
    to: "on"
condition: []
action:
  - delay:
      hours: 0
      minutes: 0
      seconds: 1
      milliseconds: 0
  - service: notify.julian_telegram
    metadata: {}
    data:
      message: test
mode: restart

Depending on how long the service takes to execute, the delay might not even be needed(?)
But with it, I can reproduce the issue 100%. Currently running core 2024.6.0.

I've definitely also seen this in some actual (light-related) automations that have the restart mode.

@bdraco You already had fixed some similar issues where script/automations cancelled in restart mode in previous releases. Are you aware of any other issues with this? (or have some time to also take a look at this? 😄 )

from core.

I have fixed a few ones recently, but this looks different. Is there a minimal reliable reproducer available?

from core.

Yeah, see the automation I have in my comment above. That's as minimal I was able to get it with reproducing consistently.
You just need to create an input boolean and test that automation.
Oh, and might wanna change the service at the end (send a notification in my example—it can do anything, doesn't really matter, as the automation doesn't get to that point when looking at the traces).

There's also one automation in the initial post.

from core.

@TheJulianJES @LucasCZE
Can you confirmed the linked PR #119100 fixes this problem?

from core.

Just FYI, I updated my HA to 2024.6.3 (the fix was included in 2024.6.2) and the issue is fixed also in my environment. 👍

from core.

Hey there @home-assistant/core, mind taking a look at this issue as it has been labeled with an integration (automation) you are listed as a code owner for? Thanks!

_{^{(message by CodeOwnersMention)}}

automation documentation
automation source
_{^{(message by IssueLinks)}}

from core.

I think I've actually seen similar behavior with restart automations in 2024.5.x and 2024.6.x.

from core.

I also see this issue with a real automation for my window blind. The automation is quite complex and I can see it almost every day. The example YAML I have included is just for the demonstration, but the issue can easily occur even in a real case of use.

from core.

Did a quick test with the example I had in my comment above and the issue seems to be fixed then.
And as always, thanks for your work!

From the automation traces, the first automation run got into the delay and was restarted then:

The second run was restarted before it even got to the delay:

And as expected, with the fix, the last run completed now:

That's just an observation, the fixed behavior seems fine to me.

from core.

Thanks for testing!

from core.

Thanks for the fix @bdraco and thanks for testing it @TheJulianJES. I really don't know how to test the fix. But as per the comment above, I believe the bug is fixed. Thank you again. 👍

from core.