Release v2.18.0

[bmachek]

What's Changed

• Ensure DB is initialized on every request.
• speedup indexing pipeline on macOS.
• Remove redundant image conversions in backend.

Full Changelog: v2.17.0...v2.18.0

---

This discussion was created from the release Release v2.18.0.

[Frank289159]

Since the backend was introduced, none of the add-on's versions have worked for me.
After the first few versions, I stopped trying. Today I decided to give it another shot, hoping it would finally work again.

It seems like a lot of interesting features have been added. Unfortunately, I still can’t use them.
When I try to start the backend, I get the following message:

The port isn’t in use elsewhere—I’ve checked that. So what am I doing wrong? Or is this just another bug that hasn’t been fixed yet?
I’m going to deactivate the add-on again and switch back to the version without the database for now. That one is still working very well.

[bmachek]

I would suppose the backend doesn't start in your case. You could create an issue for that and attach the log files. Log file for the backend is located in %LOCALAPPDATA%\LrGeniusAI\logs

[oh-codes]

This is maybe a windows only issue / the way the server is started:

@server_bp.route("/restart", methods=["POST"])
def restart():
"""
Gracefully shut down the server.
If running as an OS service (launchd/Windows Service), it will be automatically restarted.
"""
logger.info("Restart request received via API")
server_lifecycle.request_shutdown()
return jsonify({"status": "Restarting..."})

under windows with the provided installer the server is started via a.cmd - and this is not a service so it wont restart on itself.
Think the bestway woud be to have an own function to shutdown/restart in server.py

Regards

[Frank289159]

Thanks,

when I launch geniusai_server.py manually, it works.
It’s a bit of a hassle, but it’s doable. If anyone has a solution to make it start automatically, I’d appreciate it.

But I have a new problem.

After installing the search AI, which works fine, I get the following message when I launch Lightroom. It seems I can dismiss it without any further issues. But it must mean something, and always starting with an error message gives me a bad feeling.

The report file for this error contains no further information beyond what’s shown in the message.

[oh-codes]

Concerning server start see : #197

Regarding CLIP / your error Message: I ran in same issue, for my part I solved by forcing reloading the Model - maybe the download model was broken - dunno

First start server then
http:/localhost:19819/clip/dowload/start

Hope it helps,
Olivier

[Frank289159]

Thanks, Olio,
but unfortunately I can't find the APISearchindex file anywhere.

However, I was able to solve the problem with starting the server on my Windows machine much more easily. Apparently, it was simply because Windows didn't recognize the .py file extension and therefore couldn't run the file since it didn't know what to use to open it.

I just assigned .py as the file type for the Python application, and now the service starts.

BUT:
Where should I enter the http path?
My browser says: “The requested URL was not found on the server.”
Where can I find CLIP on my hard drive?
Then I'll just delete it, and then he'll have to download it again.

But apparently CLIP fixed itself, because the error isn't showing up anymore today.