I’m testing out the LLM application and it works fine to just chat with a model via Ollama. However, when I try to index a collection (using qwen3-embedding:8b via Ollama), it goes through each page in the collection and produces an error message for each page in the following format:
[task-manager-consumer] WARN o.x.c.l.i.DocumentIndexer - Error while embedding chunks [0-0] of document [xwiki:path.to.page.WebHome;]: [RequestError: 500: do embedding request: Post “http://127.0.0.1:NNNNN/embedding”: EOF]
Where the port number (NNNNN above) is a seemingly random 5 digit value that differs for each page it tries to index.
The error message is the same on the collection page as in XWiki’s logs. I’m not seeing any additional errors that might provide a clue as to what is failing or why. Does anyone have an idea as to what might be going wrong?