Hey everyone, the issue was introduced in the new version 0.8.1, which is an almost complete rewrite of the Watchman.
I tested it in prod environment before the release and it seems the problem is not reproducible on all installations, but Iām working hard to fix it. I have removed 0.8.1 and 0.8.0 from the releases for the moment.
Version 0.7.0 beta.1 should work fine. Thank you for your understanding.
@scr I also asked on GitHub, but better safe than sorry:
Iām a little unsure. I installed V0.8.1 yesterday, and after a complete system restart tonight, triggered by Proxmox Backup of the HA VM, everything ran normally and without errors.
My question: should I downgrade? Could this problem still affect me?
Not seeing any issues (yet) with this 0.8.1 (
)
But itās calling out:
notify:
- name: "email"
platform: smtp
server: "smtp.gmail.com"
as an entity-id:
Missing 1 entity(ies) from 218 found in your config:
| smtp.gmail | missing | configuration.yaml:505
even when my Ignored entities and actions list contains:
notify.email,switchā¦
Home Assistant Container Core: 2026.1.3
Version core-2026.1.3
confirmed, I went from release 6.x to new 8.x, borked the system.
I deleted altogether (so didnt do some nifty uninstall, I just deleted the folder and restarted (always first install add-ons Samba and SSH to be able to do that under any circumstanceā¦) and after that installed 7.0 beta 1.
Did have to reconfigure the options (it says migration failed somewhere, so ignore that and start all over with a fresh configuration)
I tried installing 0.7.0 beta 1 but that wonāt start due to a migration error.
Yes, delete the existing config and add a new is the MO now
Better first copy the list of items to ignore then. Things like light.turn_ because it does not understand templated actions 
Any ETA for the Non-Beta that is not broken coming out?
Yes, I did, but then decided to have a go at it and find out what it reports now.
The ignore list I had was years old so probably required updating itselfā¦
Was a quick effort to be down to nothing to report 
As always, in open source, with volunteers doing the work, the correct answer is always the same:
YESTERDAY!!!
Please be patient, letting them focus on the fix, rather than be distracted by answering when it will be out.
And also be aware that community driven development means the author may not have the time to work on it straight away. In business, organizations pay for a Service Level Agreement with response times to fix problems. This is provided free of charge, as is.
Hey guys, development version 0.8.3 is ready! I would be incredibly grateful if those who previously experienced issues could test it in their environments. I have implemented extensive safety measures to minimize any risk of freezing.
However, as this is a test build, please proceed with reasonable caution: #227
I still only see 7 beta in HACS, no pre release of 8.x? A dev build is harder to test, seeing I also use markdown cards etc. and I wonāt be able to see if it can work with my previous config. Will it confict with an existing, disabled config?
no pre release of 8.x
Not yet. Iād like to get at least 5 confirmations first (I already have two). In the past, a pre-release was accidentally pushed to general users, so I donāt want to risk that again.
A dev build is harder to test, seeing I also use markdown cards etc.
It actually isnāt. You donāt need to change anythingājust install it from the HACS custom repo and let it run to check the diagnostic sensors. Afterward, it can be safely removed without touching your existing 0.7 config.
But as always: make a backup first.
Thatās a particular sore point for me, and I plan to address it in future releases. To support the markdown card, Watchman currently saves all report data in sensor attributes, which is architecturally incorrect. I might be able to implement a proper custom card to finally get rid of this approach.
P.S. I have to say, Home Assistant handles the load quite well. A sensor with 6,000 missing entities in its attributes (result of parser stress test) doesnāt impact performance at all, which is impressive.
Watchman_dev broke things again. Simply adding the integration was enough. No chance to change the config.
Ha core restart from ssh took a very long time to get HA out of it (after I deleted the folder using Samba).
I applied the update but did not reboot. I them did something that made the existing Watchman disappear from HA. But the reboot fix was still there. So, I selected cancel or such on it. It is still there. I am concerned about rebooting my HA. Is there anything I can do before a reboot to ensure that I am safe?
Already some days ago I removed the directory from custom_components, did a reboot and since then everything is fine. And now I wait until a well tested version is ready to download.
Hi everyone, just a quick update.
Iāve been dedicating almost all my free time over the last week, including most of the weekend, to fixing this issue. I have released 10 development versions so far, and while each brings improvements, the bug is tricky because it doesnāt reproduce in every environment (including my own three setups).
Therefore, I rely heavily on feedback from those of you generous enough to test the latest dev version (0.8.3-dev10). If the issue persists, please share your debug logs - it helps immensely.
All details regarding the dev versions can be found in the GitHub tickets #222 and #227. Thank you for your patience and understanding.