Unfortunately, after the migration, all my 193 entities are renamed, so nothing works anymore. Assuming I only use 60 of them, they appear in many Node-RED scripts, helpers, screens, etc., so I would have to touch about 300 places. Obviously this is not possible, or only possible with a lot of time, is error-prone and would have to be thoroughly tested. I am not even trying to imagine how long it would take me to do this and how long my Home Assistant would be offline (including all automations, light swicthes, securities, …) This, by the way, is the recommendation that was given to me by the responsible people in the above-mentioned forum due to the lack of a migration path.
Has anyone had any experience with this and if so, what is the solution (other than the one described above)?
For me it’s a complete dead end at the moment as I don’t know how to solve the problem and I can’t upgrade anymore. The options as far as I can see are
Touch about 400 places
Simply rename 60 entities back to their pre-migration names (only 60 places to touch).
As an alternative, I’m considering replacing my 30 Homematic devices with ZigBee if I can’t find a solution to the upgrade problem.
Just remove the old integration and restart HA before installing the new integration. With that >90% will get the same entity id’s as they had before. At least if you havent touched the id’s before and they were as derived from the CCU.
Weel, in the meantime, I read your complains in the forum over there. Let me summarize and correct me, where I’m wrong:
you are not talking about a migration of the core integration here to the new custom migration
you are talking about a migration from a beta-version of the custom migration to a newer version of the custom migration
you volunteered on using the beta-version of the custom integration instead of using the core integration
you didn’t update the beta version now for over 1 year
you didn’t follow the breaking changes info now for over 1 year
you discussed this already with the more than helpful author of the custom migration
in the time of your complains you would have been able to rename these little 60 entities or even all of them multiple times, so where is the problem?
instead of these 10 minutes little effort you think about buying new hardware. Really?
To be honest, if you were already in touch with the author and co-author (and one of the maintainers of the core integration), what you are expecting here and get here?
And, you should have renamed the entities anyway already before. I did it with most of my 1.288 only from this integration. It will reduce the work if you must replace the devices and entities in the future. Because I saw, that you have still the serial number in the entity_ids, etc.
@arganto, I do not know where you got this misinformation from. Let me correct you: I don’t use beta versions. I did not volunteer using beta. I update all my stuff at least once a month. I did not see any braking changes. I am talking about changes in 300 places. Changing entities in all places is not a 10 minute task. All automations, screens, windows, motion sensors, switches, etc. have to be tested afterwards.
What I can do is change the entity names back to the original name after the migration. I am talking about 60 that I am likely to use (out of the 193). This takes 10 minutes for you, maybe 1 hour for me. And then I have to start testing all the stuff.
And this is exactly the information I am looking for, suggestions from others on what to do. Thanks for your suggestions. Let’s see if there are any more from others.
Please don’t tag users here. It’s against the forum rules.
You used a custom integration < version 1.0. I would say this is beta. You used this instead of the official core integration I linked above which is available and in HA core since 2016.
Perhaps you updated the HA core once a month and if you would have used the HM core integration this would have been updated as well. But you decided to use the custom one in such an early <v1.0 state instead. This is volunteering I would say.
No, you didn’t update the custom integration in the last 15 months. As Sukram have written the 1.0 with the listed breaking changes is available since March 2022. But you were still on the version from before March 2022 and got into trouble after updating now to a newer version (wich was necessary to run with the new HA version of course).
OK, I understand reg. taging users. I thought it was polite. I will not do it again.
But please help me with this version misunderstanding. I am using the HACS integration Homematic(IP) Local version 1.36. Are you talking about a different version or integration? I am trying to find out with this question what I am doing wrong.
There are two integrations. The official one since 2016
and the Custom HACS one (currently in version 1.37.6). And your recently followed a migration guide from HACS component < v1.0 to HACS component >=1.0. But 1.0 is available sind March 2022 and the guide is from September 2022. So you were not up to date.
I think the HACS integration is better. But you have decided to use a very early version of this instead of the official one above. Of course, this is no help for you now, but I would just update the entity_ids quickly once and fine.
That you just used a beta version which generated the id’s different than the core integration of the HACS integration is now generating or whatever went wrong 1,5 years ago is is of course a pity, but really an edge case. Does not help you here but is the risk of testing early components.
I am beginning to understand. I remember when I was trying to integrate Homematic I was warned not to use the HomeMatic add-on and told to use “RaspberryMatic CCU” (and “Homematic(IP) Local”) instead. What I read from your message is that I used it too early (probably in beta) and that’s why I have these strange entity names and can’t migrate easily.
I have to decide whether to just change the entity names back to what they are today OR to change all my flows, screens, etc.
Sometimes I think I have too much SmartHome and should probably reduce the functionality - which is what my wife wants anyway
Thanks for your support.
Replace all the Homematic products.
These products were carried over from my old Telekom Magenta SmartHome integration. With a new installation I would try to save integrations and probably not integrate Homematic. On the other hand, the Homematic products are stable, seem robust and come from a German company.
Change all affected devices to the old device name.
It turned out that of the 30 devices with 193 entities, I was only using 68 and was able to reduce this number to 50.
Change all Node-RED flows, screens, wizards, automations and configurations (e.g. in configuration.yaml).
I finally did this step and there were 204 changes.
Procedure and time needed
I spent a little over a month, from the end of May to the end of June, trying to understand the problem. It took about 8 hours.
On 01/07/2023 I started the analysis. I found out how many places I had to change: 50 entities, 107 times in Node-RED, 15 times in helpers, 64 times in screens (the number should increase during QA), 3 times in configuration.yaml. The analysis was finished on 02/07/2023. It took 5 hours and 20 minutes.
I started making changes in the late afternoon of 02/07/2022. This took 3 hours and 15 minutes. In the process, I also discovered what I had overlooked in my analysis, namely the configuration.yaml, which also had to be changed in three places (and these changes are still not working today, I don’t know why).
Early on 03/07/2023 more bugs were found. 15 device nodes in Node-RED had lost their values and had to be repaired. One water sensor had a different name and had to be changed in Node-RED, Screens and configuration.yaml. All three water sensors return a different status, lower case instead of upper case. This meant that configuration.yaml had to be corrected in three places and Node-RED in four. Then I discovered wrong values in screens that I had overlooked in my analysis. Time spent: 3 hours.
On 03/07/2022, after all upgrades, everything is running since the afternoon. Only the translation of the status of the water sensors from English to German in the configuration.yaml does not work for reasons unknown so far.
All in all, this upgrade took about 6 weeks. The time required was 19.5 hours. It is a pity that there was no migration path that could have been run automatically.
Hi SukramJ, please point me to the release nodes so I can see what I have not read and missed. I usually read the release nodes (change logs) that HA provides before I update anything important. I would really like to know where I made a mistake, where I missed something, so that I can be even sharper next time.
I really respect the contribution you and others make, I have done the same in the past (though not for HA).
