Search Index and AI Problem

[FasteryXD]

Hello,
I installed Gramp for the first time today and I'm very satisfied so far. They've done a really great job. Unfortunately, I'm running into two problems that I can't solve on my own. I've been trying for hours now and nothing has worked.
The first problem was that after I created and started a Docker Compose on Ubuntu and then created the first person with a location, the error message "search index needs to be rebuilt" appeared at the bottom. Then I went to the settings via the link and apparently rebuilt everything, which the green checkmarks seem to indicate, but the message didn't go away, and the red crosses at the top didn't go away either.

The second problem is the AI. I created Ollama, which works wonderfully with Nextcloud and Home Assistant. So, I've filled out everything so far and have now received the chat, but it said "Invalid message format." After that, I tried to change a few settings and got the following error message: Chat API error detected. After I undid the "improvements", I kept getting the error message Chat Api error decodered instead of Invalid message format, even though I had reset everything to "initial".

  GRAMPSWEB_LLM_Base_URL: "http://xxxx/v1"
  GRAMPSWEB_VECTOR_EMBEDDING_MODEL: "sentence-transformers/multi-qa-mpnet-base-dot-v1"
  GRAMPSWEB_LLM_MODEL: "llama3.2"
  OPENAI_API_KEY: "ollama"
  GRAMPSWEB_LLM_MAX_CONTEXT_LENGTH: "5000"

Please help!

[DavidMStraub]

Hi, please share your full docker compose (with secrets redacted).

[FasteryXD]

Like what? Can you explain?
I haven't been on Github that often or for that long.

[DavidMStraub]

https://docs.github.com/en/get-started/writing-on-github/working-with-advanced-formatting/creating-and-highlighting-code-blocks#syntax-highlighting

[FasteryXD]

Ok, I got it, the yaml file is now above.

[DavidMStraub]

/app/index → /app/indexdir 🙄

[FasteryXD]

Sorry to bore you, but yes, that solved the problem with the search index.

But I'm not getting any further with the AI (Ollama). The GRAMPSWEB_LLM_Base_URL is 100% correct because it works under Nextcloud and Home Assistant.

But I get the following error message:

INFO:httpx:HTTP Request: POST https://api.openai.com/v1/chat/completions "HTTP/1.1 401 Unauthorized"

And this is totally confusing me, so I'd like to ask for your advice again.

[DavidMStraub]

Sorry to bore you, but yes, that solved the problem with the search index.

I ensure you I am not bored. Sorry for my undiplomatic emoji, but sometimes I am just puzzled how spending time on a sunny day to help somebody find the typos in their config files is not even worth a "thank you".

But I'm not getting any further with the AI (Ollama). The GRAMPSWEB_LLM_Base_URL is 100% correct because it works under Nextcloud and Home Assistant.

Sorry, I cannot support with the Ollama problem (because I don't have a working setup to compare to and no time to set one up), but if you identify the problem and find a gap in the documentation, a fix would be very welcome. I suggest to look at the source code in Gramps Web API and try to call your local Ollama with the same library to debug.

https://github.com/gramps-project/gramps-web-api/blob/master/gramps_webapi/api/llm/__init__.py

[FasteryXD]

I know that in today's society, saying "please" and "thank you" is becoming increasingly rare, which makes it all the more important to maintain such customs. I wanted to first get to the bottom of the problem with you and then express my sincere gratitude!
Ganz lieben herzlichen Dank!!!👍

I'll let you know when I can resolve the issue. You haven't set up a Discord server, have you??

[DavidMStraub]

Nope, no Discord unfortunately...