After Update to Version v4.4.8 error

[Kieft-C]

home-assistant_solcast_solar_2025-11-20T19-50-38.963Z.log

For any reason my hole Solcast not work anymore.
I can't go back or install new. I allways get failures, but got no idea, what happens.
Any help idea here? maybe I did something wrong.

[gcoan]

@Kieft-C can you provide details of what the files are in /config/solcast-solar directory, what names and are they containing content or empty?

At the moment the integration is reporting

2025-11-20 20:49:11.627 WARNING (MainThread) [custom_components.solcast_solar.solcastapi] There is no solcast.json to load, so fetching solar forecast, including past forecasts
2025-11-20 20:49:11.629 WARNING (MainThread) [custom_components.solcast_solar.solcastapi] API polling limit exhausted, not getting forecast for site 94ee-ba16-d8c2-d400, API used is 10/10

So it can't download any data from Solcast until after midnight UTC.

Wait until then and see what the integration does, it looks like its trying to download your entire sites info but can't right now because you're out of API calls. Waiting until after midnight and it could well resurrect itself

[Kieft-C]

both containing some lines.
Thx for very fast reply. Looking forward for the next API Call and see what happens. :)

[gcoan]

OK thanks, so the integration has moved those there from /config

Can you look in /config and see if there are any files named 'solcast', in particular solcast.json as that's the one the integration is complaining it can't find so has to go and do a download (but can't):

There is no solcast.json to load, so fetching solar forecast, including past forecasts

If there is a solcast.json file or indeed any other solcast files in /config, can you move them to the solcast-solar subdirectory which might well resurrect your forecast

[BJReplay]

Thanks @gcoan for your initial assistance troubleshooting.

@Kieft-C it looks like you have had at least two restarts.

The log you have provided (Thanks!) indicates that the files that the integration expects are in the /config/solcast_solar/ folder - that tells us that you are running v4.4.8.

What it doesn't tell us is whether you were running an earlier version, but your title of this issue suggests that you were.

So, here's what appears to have happened:

• You were running v4.4.7
• You upgraded to v4.4.8
• v4.4.8 moved the files to the new config folder at /config/solcast_solar/
• For some reason - this is not clear from the logs, or from your issue report, you have restarted again.
• The log you attached is from this (or a later restart).
• It tells us that it detected a prior crash.
• The log you attached indicates that the files that are in that directory are zero length. That should not happen, and is a concern.
• You've run out of API calls for the day, so the integration can't get the data it needs to start. You're stuck, for the moment.

Can you tell us a bit more about why there were multiple restarts, and what you recall about why things stopped working?

Do you have any older log files? If so, please attach.

Now, as to recovery:

• Just wait - in about 1 hour and 40 minutes, you'll get your next batch of API calls, and the integration should be able to initialise, or
• If you have a backup from before the upgrade, if you copy the files solcast*.json file from your /config/ folder from the backup to the /config/solcast_solar/, you should be able to restart, and it should use those files.

[autoSteve]

I don't know what on Earth you did to get in this situation. You don't provide much detail, @Kieft-C .

Please show the /config/ folder full contents (a.k.a /homeassistant/). Please show /config/solcast_solar/ folder full contents. I cannot fathom how you could have broken this just by upgrading.

Somewhere in your log history should be some lines that look like "Migrating config directory file %s to %s", and I want to know whether any exception occurred around this time.

Report back if this resurrects itself.

[Kieft-C]

Thx for all your work and help.

After the first API Call, it work again normal.

But then maybe u can add a Warnung what give a Info that u have to wait until a new api call or even when u have Allready used all calls the day that u have to wait until next day? That would be helpful.

And how I got in this ? I got now idea, is only because of the missing api Call?
And Im Sorry to give not more Details, I got not more. It works almost a half a year without Problem incl all Updates. Yesterday was the First time with a error here. :)

[drulia]

I got similar issue where after updating to v4.4.8 it stopped working. It was saying initially that my config/solcast.json is incomplete, even though it is actually in config/solcast_solar folder.
Logs have this and then many more errors, which presumably all stem from this incompatible solcast.json:

2025-11-20 17:19:14.544 CRITICAL (MainThread) [custom_components.solcast_solar.solcastapi] The solcast.json appears incompatible, so cannot upgrade it
2025-11-20 17:19:14.563 CRITICAL (MainThread) [custom_components.solcast_solar.solcastapi] The solcast.json appears incompatible, so cannot upgrade it
2025-11-20 17:19:14.857 ERROR (MainThread) [homeassistant.config_entries] Error setting up entry Solcast Solar for solcast_solar: init_incompatible
Traceback (most recent call last):
  File "/usr/src/homeassistant/homeassistant/config_entries.py", line 761, in __async_setup_with_context
    result = await component.async_setup_entry(hass, self)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/config/custom_components/solcast_solar/__init__.py", line 433, in async_setup_entry
    raise ConfigEntryError(translation_domain=DOMAIN, translation_key=TRANSLATE_INIT_INCOMPATIBLE)
homeassistant.exceptions.ConfigEntryError: init_incompatible

inside config/solcast_solar there are these files:

solcast.json
solcast-actuals.json
solcast-actuals-dampened.json
solcast-sites.json
solcast-undampened.json
solcast-usage.json

[BJReplay]

@drulia any chance you can attach the home-assistant.log and home-assistant.log.1 so we can see as much as possible of the log?

[autoSteve]

And then what happened, @drulia?

This is fascinating, given over 4,800 downloads so far. The Internet should have exploded if the code had a serious issue. Surely 4,800+ people have not fixed an issue by themselves or not noticed?

The check for incompatible data happens in load_saved_data(). The exception is raised only in the event that the file can be read, has a JSON structure, is a version different to that expected (or a default of 1), yet the keys "forecasts" and/or "site_info" are not in the JSON dictionary contained in the file, indicating a corruption that has blasted these keys.

You state which files are there, but not what is inside solcast.json. Without that I am diagnosing a symptom blind, and also without context.

You probably don't have DEBUG logging available, so would not see other potentially related messages, and neither will I ever.

And you probably should have created a separate issue because your symptom might match the OP (won't start after upgrade), but your circumstance sounds TOTALLY different. "Me too" just helps to mask the issue, but thanks for the additional logged detail.

[autoSteve]

attach the home-assistant.log and home-assistant.log.1

Won't exist if you're on the latest Home Assistant version, @drulia. Log download under the Settings | System | Logs section for the win. This will only get the equivalent of home-assistant.log and not .log.1.

I love and hate at the same time that HA made this change to remove the log files.

[autoSteve]

And how I got in this ? I got now idea, is only because of the missing api Call?

What did you get, @Kieft-C? Your post is unclear.

[autoSteve]

Ahhh. You mean, "How did this happen?", @Kieft-C.

I have no idea yet, and almost no data to analyse.

[BJReplay]

And how I got in this ? I got now idea, is only because of the missing api Call? And Im Sorry to give not more Details, I got not more.

@Kieft-C

OK, so, to be clear, you have no idea what you did, can't give us any more details, but you now want us to tell you what happened?

This is not how things work.

We have no idea.

The more information you can provide, the more chance we can work out what happened.

Given that you can't provide any more information, we can't tell you what happened.

It is working.

Be happy, and remember to check that your backups are working.

[autoSteve]

It was saying initially that my config/solcast.json is incomplete, even though it is actually in config/solcast_solar folder.

Please clarify or amend this statement, @drulia. It is unhelpful at best, and misleading at worst.

That is not what the supplied log says. It mentions nothing like that.

Those logged messages supplied do not mention /config/solcast.json anywhere. I checked your message four times.

It seems unlikely that you are "upgrading" from a positively ancient and incompatible integration, yet I have no context, and I'm bad with names. You might be. That you are One in 5,112 now to have hit this strengthens my thought that you may be "migrating" from a different integration.

[autoSteve]

I am a whisker away from closing this as unplanned, @BJReplay.

It is muddled with two crossed streams of completely different information with almost zero information to analyse from either.

5,177 downloads now without all Hell being raised.

• @Kieft-C, if you give us more, we can analyse it. Giving us more may help others.
• @drulia, if you give us anything other than "I upgraded, then hit this message", we can analyse it. Like context. Like solcast.json. Like a full log. Like what you "upgraded" from. Like your HA version.

[Kieft-C]

@autoSteve thanks and totally understand.

I update from 4.4.7 to 4.4.8 like always.
Doin nothing special, restart (soft) HA after the update from the integration and until then I get several problems.
Normally I jump then back to the previous version but this also not work.
After that I read in the info, that the rollback is not easy possible and then I deinstall everything and restart with a fresh 4.4.8 and start debug mode, with the log which I send here.

And then get the good feedback to wait here until the next API Call is possible, and it works.

So no bad vibe from my side, because I guess it is my own fault, but can't recreate it.

[autoSteve]

That sounds like a bad day, @Kieft-C. Sorry.

Yeah, we moved the cache files from /config/ to /config/solcast_solar/ as the release note said.

The really fascinating thing for me is that you hit any problem at all after the re-start.

One of the first things the code does now is to migrate any cache from /config/, and almost in the same breath after it establishes the configuration folder location (and creates it if it does not yet exist prior to migrating anything).

The problems that you hit that resulted in initiating a roll-back may be important.

The Home Assistant "detailed" log retains history that crosses restart events. You should be able to scroll back to when you upgraded. Settings | System | Logs then "Show full logs" at the three-dot menu top right of screen.

[BJReplay]

I am a whisker away from closing this as unplanned

I was there a while ago, too, @autoSteve

I'm there now.

I'm going to convert this to a discussion. That way, if @Kieft-C has more information, it can be provided, there. This isn't an issue - @Kieft-C has an operational system, and we are unlikely to get any more information to tell us what happened so we're unlikely to be able to explain why it happened.

If @drulia can provide more information - again unlikely - including version upgrading from - and whether or not it was this integration, then they can either create new discussion, or if it is actually an issue, then follow the instructions and create an issue. I'm going add notes to the issue instructions to say that me-too issues are banned, for obvious reasons. Converting to discussion now.

[autoSteve]

@drulia, if you actually are trying to migrate from an incompatible integration (i.e not BJReplay / Oziee) then the procedure involves removing all solcast*.json files and starting again, and is not an "upgrade" just because two Home Assistant things say "Solcast".

If you're coming from a BJReplay release then I'm 'all ears' for further information.

[autoSteve]

Some good has come of this.

• The 'incompatible' state is not tested for if the subject file is not a JSON dictionary, basically now stating bluntly that the file is corrupt.
• The 'incompatible' error raised now explicitly states the folder that the inspected file is in, and its tested name, not a blanket solcast.json, which might be misleading.
• The CRITICAL message is only logged once. We don't like CRITICAL messages...
• The code is more efficient in pointing out the explosion and getting back to a repeatedly "Initialising" integration state because broken.

It is remotely possible, but unlikely, @drulia that you hit #392, which killed your solcast.json making it zero bytes long on shutdown. If this did happen then the irony is that v4.4.8 should fix this. But again, I am guessing without any detail.

[bigtomi]

Witam mam to samo po aktualizacji 4.4.8 błąd jak mam to naprawić pisze o tym Przenieś wszystkie pliki pamięci podręcznej i konfiguracyjne config/solcast_solardo@autoSteve ale jak to przenieś czy taki folder będzie co mam zrobić

[BJReplay]

wonder whether I should not code up a 'repairable' issue raised on this kind of massive corruption occurring, on the back of refusal to provide detail

For 1 (or, being generous, 3) in 10,600, who almost certainly hit the combination of

• 429 storm
• version change
• at least one added sites or did a sacrificial API key change
• at least one, when pressed for evidence, provided files from different versions with different sets of sites updated more than a week apart.

Yeah, nah.

That said, if you are truly bored @autoSteve, knock yourself out, but imagine the support calls for people who can't read an repairable issue that says "this will delete all your files".

Other considerations that can't easily be warned, and less easily comprehended in the issue message include utilisation of API calls, potential exhaustion of API calls (possibly part way through recovery for multiple sites) and/or API calls (and you can't know if the use has used force updates) and does that lead to deferred recovery?

Yeah, nah.

[BJReplay]

I reckon close / lock, @autoSteve?

Radio silence for the last couple of days.

[autoSteve]

I would assert, "yes", BJ.

If someone has further grievance, then, "Me too" here is not going to help anyone, nor does 'radio silence', nor was this a particularly enlightening discussion.

We still have about ~6k estimated un-upgraded users, so further sh*t might pop up if there be another 429 storm and subsequent chicken-season is declared. Having solcast.json smashed is rare, unless one performs (an alliterous) repeated ritual ridiculous restart after non-sensically changing an API key, which raises the odds on pre v4.4.8.

Me being the occasional betting man, I'd place the odds at around 100:1 for that.

I initially thought that a lock would be extreme, but yeah, nah. That is warranted on reading back through this; it should never come back.

I have also bolstered the start-up corruption checking code, but will only proffer that as a PR once @Nilogax has had his turn. My code is non-critical, and quite conflicty in places, so I won't go there just yet. I will also squeeze into that subsequent PR the pytest code for his code to bring that back to 100% coverage.

I'm not bored yet. Lately I've bent my head back into C++ mode wrangling some ESPhome goodness.

(The Kaiser Chiefs were on form last night... Riot.)

[BJReplay]

I've bent my head back into C++ mode wrangling some ESPhome goodness.

I have only done ESPhome in via ESPhome builder, which is cheating.

The Kaiser Chiefs were on form last night... Riot.

Just grabbed tickets to Lloyd Cole for a few days before Sakura season trip next year

[autoSteve]

A good 'commotion' to have... ❤️

[drulia]

okay, I was upgrading by clicking on notification saying that there is a new version, I did not reinstall anything, nor I had previously other version. Installed from hacs originally.

I'm using latest 2025.11.2 version of HA.

initial message about config/solcast.json is incomplete was shown in UI, I remember recently seing solcast files in config dir, but now all those files are in config/solcast_solar folder.
Here is the message:

Now solcast.json does contain two forecasts but do not have site_info, it's a large file, but I've attached it. I can see siteinfo in and forecasts in three other files, solcast-actuals.json, solcast-actuals-dampened.json and solcast-undampened.json, these look pretty identical on the first glance, I'll attach one as an example.

solcast.json
solcast-actuals.json

After previously posted log, there are quite a few of these errors following:

2025-11-20 17:19:16.116 ERROR (MainThread) [homeassistant.components.websocket_api.http.connection] [139881141024736] Error handling message: Unknown error (unknown_error) Giedrius from 192.168.0.135 (Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:145.0) Gecko/20100101 Firefox/145.0)
Traceback (most recent call last):
  File "/usr/src/homeassistant/homeassistant/components/websocket_api/decorators.py", line 28, in _handle_async_response
    await func(hass, connection, msg)
  File "/usr/src/homeassistant/homeassistant/components/energy/websocket_api.py", line 99, in with_manager
    await result
  File "/usr/src/homeassistant/homeassistant/components/energy/websocket_api.py", line 231, in ws_solar_forecast
    forecast = await forecast_platforms[config_entry.domain](hass, config_entry_id)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/config/custom_components/solcast_solar/energy.py", line 36, in async_get_solar_forecast
    or (coordinator := entry.runtime_data.coordinator) is None
                       ^^^^^^^^^^^^^^^^^^
AttributeError: 'ConfigEntry' object has no attribute 'runtime_data'

These are identical and happened at least 15 times in a span of a couple of seconds. Then there is nothing else until today, (edit) which likely happened when I tried to reload integration (end of edit):

2025-11-21 08:58:23.601 WARNING (MainThread) [custom_components.solcast_solar.solcastapi] Get sites failed, last call result: 999/Prior crash, using cached data
2025-11-21 08:58:23.605 WARNING (MainThread) [custom_components.solcast_solar.solcastapi] Resetting usage for ******LtnKGh, last reset was more than 24-hours ago
2025-11-21 08:58:23.833 CRITICAL (MainThread) [custom_components.solcast_solar.solcastapi] The solcast.json appears incompatible, so cannot upgrade it
2025-11-21 08:58:23.837 CRITICAL (MainThread) [custom_components.solcast_solar.solcastapi] The solcast.json appears incompatible, so cannot upgrade it
2025-11-21 08:58:24.123 ERROR (MainThread) [homeassistant.config_entries] Error setting up entry Solcast Solar for solcast_solar: Cached `config/solcast.json` data incompatible. Delete it
Traceback (most recent call last):
  File "/usr/src/homeassistant/homeassistant/config_entries.py", line 761, in __async_setup_with_context
    result = await component.async_setup_entry(hass, self)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/config/custom_components/solcast_solar/__init__.py", line 433, in async_setup_entry
    raise ConfigEntryError(translation_domain=DOMAIN, translation_key=TRANSLATE_INIT_INCOMPATIBLE)
homeassistant.exceptions.ConfigEntryError: Cached `config/solcast.json` data incompatible. Delete it

Let me know if there is any other info I can provide.

[Kieft-C]

@drulia this UI error Message is exactly what I also see yesterday after updating.

[BJReplay]

OK, @drulia the siteinfo is in solcast.json:

You have attached a file that was last updated a day or so ago, for two sites.

You have also attached a solcast-actuals file that was updated almost two weeks ago for a single site:

You're leaving out a lot of information.

But, now we're at pure guesswork stage, here.

You've changed API keys or added a site. Possibly while trying to recover from the 429 storm.

You've got a mix of version 7 and version 8 schema sites.

solcast.json is already at version 8, so the integration says "I can't upgrade it", but other files are at version 7, and for not as many sites.

You're expecting us to understand and explain how to fix things, without providing us with nearly enough information.

How many sites to you have.
Which version are up upgrading from
When did you upgrade?
When did you last fetch data?
Have you provided the attached files from the same version at the same time? (No, obviously not, given the differing last-updated timestamps).

To be honest, your best bet is to delete the offending file - as instructed by the integration - and let it start again.

Try following instructions.

@Kieft-C - same message for you.

If you want to actually tell us everything that you did that lead to this, then we've got a chance of telling other unlucky people what went wrong, or we've got a chance of finding a bug and fixing it, but by withholding information, we have little chance of doing either.

[autoSteve]

So... The UI message about config/solcast.json is generic on failed start, and will be fixed in a subsequent release. The strings.json and translations will in future carry a placeholder to insert the exact file name where the problem is encountered. That release will identify exactly the file that is corrupt, and where it is located.

But this is not an admission of failure of the integration. I find no relevant fault.

This should solve as concrete evidence for multiple files offered as 'evidence' of failure, spanning weeks with differing sites.

The file name is also logged. Arguably this log message was not wrong. Suffice to say the logged detail in this issue/discussion has been slim at best, hence my not bothering to check further. It logs well today, and I don't think this has changed.

if self.status == SolcastApiStatus.DATA_INCOMPATIBLE:
    _LOGGER.critical("The %s appears incompatible, so cannot upgrade it", filename)

CRITICAL is hard to miss. The file name along with its folder should be logged.

The thing that bothers me here is the logging offered up in the context of...

self._filename_actuals = f"{file_path.parent / file_path.stem}-actuals{file_path.suffix}"

This should produce a file name of the parent directory, as well as the actual file name. It does for me.

Data cache /config/solcast_solar/solcast-actuals.json exists, file type is <class 'dict'>

It didn't for you, @drulia, for The solcast.json appears incompatible, so cannot upgrade it, which should have been a fully-qualified pathname, including the folder for solcast.json...

In this circumstance, the file that is corrupt is certainly solcast.json (but it is not at config/solcast.json because it will have moved to config/solcast_solar/solcast.json). The UI message just does not say precicely where it is.

The file will still be corrupt, so that is what it is.

And the message to point this out will now be fixed in a coming release.

And still I scratch my head about why the folder was missing in the message...

[autoSteve]

This thread was not answered by @drulia, @Kieft-C. This thread was further confused by @drulia. It irks me that he is now noted as "most helpful" for the last 30 days in the discussions.

The answer here is, "Take off, and nuke the site from orbit", meaning remove all solcast*.json files and restart HA.

It is the only advice we can give, given the vacuum of quality information.

You will lose history.

Either that, or restore from a backup and re-attempt the upgrade, noting every circumstance, every log message, and saving every file.

7,261 other happy customers says the code is not broken.

[BJReplay]

I have used my powers for ☯️