In the previous post, I tried to send Chroma telemetry data directly to Honeycomb. There are a lot of combinations of configuration that I tried but nothing works. Now I will use OpenTelemetry Collector to act as intermediary between Chroma and Honeycomb.

The Setup

I use docker compose for this. The docker-compose.yaml file's content:

services:
  otel-collector:
    image: otel/opentelemetry-collector-contrib:0.111.0
    platform: linux/amd64
    command: ["--config=/etc/otel-collector-config.yaml"]
    volumes:
      - ${PWD}/otel-collector-config.yaml:/etc/otel-collector-config.yaml
    networks:
      - internal
  server:
    image: chromadb/chroma
    volumes:
      - chroma_data:/data
    ports:
      - "8000:8000"
    networks:
      - internal
    environment:
      - CHROMA_OPEN_TELEMETRY__ENDPOINT=http://otel-collector:4317/
      - CHROMA_OPEN_TELEMETRY__SERVICE_NAME=chroma
    depends_on:
      - otel-collector

networks:
  internal:

volumes:
  chroma_data:

The original source of the file is the docker compose file provided in Chroma docs. The example uses Zipkin. I deleted it since I am going to use Honeycomb.

In the compose file, there is a reference to mount a yaml file, otel-collector-config.yaml, in otel-collector container. This is to configure the OpenTelemetry collector. The content of the file is as below.

receivers:
  otlp:
    protocols:
      grpc:
        endpoint: 0.0.0.0:4317
      http:
        endpoint: 0.0.0.0:4318

exporters:
  debug:
  otlphttp:
    endpoint: "https://api.honeycomb.io:443"
    headers:
      "x-honeycomb-team": "secret-honeycomb-api-key"

service:
  pipelines:
    traces:
      receivers: [otlp]
      exporters: [otlphttp, debug]
    metrics:
      receivers: [otlp]
      exporters: [otlphttp, debug]

Check if both chroma and OpenTelemetry collector containers run as expected.

docker compose ps

The test

To test that this setup works, change to a new directory and create a new Python virtual environment inside the directory.

python3 -m venv venv
source ./venv/bin/activate

Install chromadb python package.

pip install chromadb

Using Python REPL, use chromadb client to connect to chromadb container.

import chromadb
client = chromadb.HttpClient(host='localhost', port=8000)

The act of connecting to chromadb running inside the container using chromadb Python package will send traces and metrics to Honeycomb.