This one is very strange - after clearing the cache (due to an unrelated Shelly issue) I noticed that all of my home assistant auto discovered devices have this error
I have tasmota devices also through mqtt which do not have any issue .
I tried installing and reinstalling the home assistant binding (which was working fine today until I deleted the folder temp and cache) tried a bundle:diag but couldn’t really see anything obvious. I have python scripting installed too, is there any another dependency home assistant outside of mqtt?
I use docker.
So i tried updating to the snapshot, tried the latest version, and it’s always the homeassistant auto discovered mqtt things that get the error.
My tasmota things don’t get auto discovered and they are found correctly.
So right now I have about 30 things that are complaing about the home assistant binding:
Just guessing, but “waiting” is the problem. I know that the HA binding and the MQTT binding were separated not long ago. Maybe there was some script that ran in the background when that happened that was affected. Did you try deleting both, then adding them back?
EDIT: Also recall some HA issue with naming. See from the picture 27 things in the inbox. Is that normal?
About the 27 things, yeah those are matter things discovered from the Shelly matter plugin. They are unrelated.
About the separation of the mqtt and the homeassistant I am aware and it did work for quite a while back until the cache and tmp deletion… so I’m at a loss.
Mqtt is working though..
Alright let’s try it.
First removed the mqtt one - the home assistant binding was also removed together…
Curious. Ok, now adding the mqtt binding …
Done.
Now the home assistant one..
And….
Same thing
Status:
One more question after all. Have you restarted your container without the clear-cache? I have found that after an upgrade (clear-cache by default) that something doesn’t work (in my case IP cameras). So, I wait about 10 minutes and docker restart openhab-openhab-1. I never clear-cache voluntarily.
On the subject of deleting the folders or the clear-cache command, I believe they do exactly the same thing, so do not think that is related to your issue.
@Pedro_Liberal How are your homeassistant MQTT discovery entries looking? Are they the previously discovered entries from their creation date with the MQTT broker running separately, or might they have been newly rediscovered with slightly different definitions?
Which might be the newly discovered entries in your Inbox - 27 = roughly 30
Hi! No, those 27 results are from a matter discovery, and they are my Shelly devices. The home assistant things I have were working with the home assistant binding previously until I deleted the cache and tmp folder.
I’ve been thinking about my workflow and initially I recreated them through my windows machine, which might have messed up with the permissions. Since then I also deleted them outright and restarted the container but I still have the same issue.
This one is a little frustrating ..
Yep I’ve restarted the container several times since then. I left it the entire day today by itself hoping it would sort itself out but 12 hours later and they are still in handler missing error.. @ccutrer sorry to ping you! do I remember correctly you worked on this binding? Do you happen to have any hints?
Can you run bundle:status org.openhab.binding.homeassistant and bundle:diag org.openhab.binding.homeassistant from the CLI? The waiting is definitely the problem, but we need to figure out why it’s waiting. It’s not unusual for the initial install to need an extra restart of openHAB because of how Graal finds script engines and not playing well with OSGi and dynamically installing more bundles in the wrong order. Each time you restart your docker container you’re not re-creating it from scratch, right?
Apologies for the hijack, but if you are in the Matter landscape, you should try the Matterbridge add-on in HA. Much like OH Matter binding can expose non matter devices to other controllers, Matterbridge does the same. Some channels may be inconsistent (as they adhere to the Matter standard) but 90%+ works fine. Anything else I use webhooks or direct to HA REST with HTTP binding.
I think many are playing too fast and loose with the cache clearing (not that it explains why this won’t start working again). Normally, clearing caches are pretty harmless, they will repopulate themselves automatically, but in this case we’re talking about the OSGi cache - which holds all the bundles installed in the system. So, when you clear this, everything must actually be reinstalled at next startup.
Because things depend on each other in an often complex way, that’s often not a painless process, and it can take a restart or two to make everything install again (because some things get stuck because they depend on other things that aren’t there). Ideally, this would always resolve itself, but it’s not always the case.
If I want to make sure that a bundle is “refreshed”, I’d rather use bundle:update than doing a full cache clearing.
