DST bug in History Graphs?

Tonight at 03:00 DST (summertime) was switched to normal time (winter time) in my country, with the clock jumping to 02:00.
It looks like that since that moment all my History Graphs show frozen values.
This are some examples:



The sensor values on which the history graphs are based are still showing correct values:



The HA system is a HA Supervised set-up:

  • Debian 11 on AMD E350 APU x86-64
  • Home Assistant 2023.10.5
  • Supervisor 2023.10.1
  • Frontend 20231005.0 – latest
  • MariaDB version: 2.6.1 with 30 days of history days and a default 5 seconds commit interval

A possibly related log entry could be:

Logger: homeassistant.components.recorder.core
Source: components/recorder/core.py:380
Integration: Recorder (documentation, issues)
First occurred: 02:46:36 (1 occurrences)
Last logged: 02:46:36

The recorder backlog queue reached the maximum size of 100081 events; usually, the system is CPU bound, I/O bound, or the database is corrupt due to a disk problem; The recorder will stop recording events to avoid running out of memory

Another, but possibly related problem that I see is that some Scrape sensors are since tonight suddenly showing strange wrong values, while the source websites are still showing normal values.
This are some examples:

And this is from the source website:

EDIT: No, this Scrape sensor problem appears to be caused by the source website being modified, coincidentally also last night. So this problem is solved now, but the problem with the History still remains.

EDIT2: After rebooting the system everything reverted back to normal again, so no database reset was needed.

Is this indeed caused by a DST bug, or is it coincidence that this happened tonight?
Is this related to the MariaDB database?

Is anybody else seeing the same?
And what is the best way to solve this: do I have to reset the database?

My guess is that it is a visual bug.
The database in HA should use UTC, which is not affected by DST, but when trying to render the graph in you view, then it tries to convert the values and then the hour for DST switching will be present twice.
It should only be the rendering that is affected then, not the actual database.

Thanks, this is an interesting thought.
But then how does this relate to the log entry about the recorder being stopped?

Anyway, in the mean time I rebooted the system, and since then everything reverted to normal again:



So luckily no reset of the database was needed.
Still I would be interested to know whether others have been experiencing the same thing right after the DST swap. So anybody?


In my graphs it just shows 2 o’clock double. Everything kept working as expected.

Home Assistant 2023.10.5
Supervisor 2023.10.1
Operating System 11.1
Frontend 20231005.0 - latest
default database with 40 day history

Thanks, I am seeing the same, so that’s OK.

Since no one else is reporting similar problems like I had, I assume it was not a DST bug but just an inexplicable glitch, that coincidentally happened right after the DST swap.