Self-hosting doc and ingestion issues

Overview

I am working on getting a simple test to work with self-hosting setup via the docker route using the code pulled from the repo today (Jan 28 2025)

Test

As easy as I could imagine, in the chat adding "I like peanut butter, add this to memory" as well as another simple phrase, second case including a label.

Problems

API key

In the docs the OPENAI_API_KEY is listed as optional, but I had to add one based on to get anywhere (error on the api key in the logs.) Is this intended? I would rather use my anthropic key if possible.

Ingestion fail

After adding the key, i tried again and now the chat responds that the memory is added and an "ingesting" spinner appears, but it just spins. The log prints ongoing seemingly successful notes.

core-app  | GET /api/v1/ingestion-queue/status.data?_routes=routes%2Fapi.v1.ingestion-queue.status 200 - - 3.528 ms
core-app  | GET /api/v1/ingestion-queue/status.data?_routes=routes%2Fapi.v1.ingestion-queue.status 200 - - 3.496 ms
core-app  | GET /api/v1/ingestion-queue/status.data?_routes=routes%2Fapi.v1.ingestion-queue.status 200 - - 3.837 ms
core-app  | GET /api/v1/ingestion-queue/status.data?_routes=routes%2Fapi.v1.ingestion-queue.status 200 - - 3.596 ms

I also selected "retry" from the documents tab on each of these.

All said and done there is NOTHING actually ingested (memories are empty) and this all consumed 44 requests and ~90k tokens on OpenAI.

What to do

I don't know where to start trouble shooting. Somewhere else in the docs I think it mentioned some versions are tagged for self-hosting, so maybe 0.3 is not ready? I did not see any such tags though.

LMK what other details I can provide.