qdrant exporter

Metrics

everything the exporter publishes at /metrics.

endpoint

once the exporter is running, metrics are scrapeable at http://localhost:9999/metrics.

sample output
# HELP qdrant_up Whether Qdrant is reachable
# TYPE qdrant_up gauge
qdrant_up 1

# HELP qdrant_collection_points Number of points per collection
# TYPE qdrant_collection_points gauge
qdrant_collection_points{collection="my_collection"} 150000
qdrant_collection_points{collection="test_vectors"} 50000

available metrics

metricdescriptionlabels / values
qdrant_upwhether the exporter can reach qdrant1 = up, 0 = down
qdrant_collection_pointsnumber of points in each collectioncollection
qdrant_collection_vectorsnumber of vectors in each collectioncollection
qdrant_collection_indexed_vectorsnumber of indexed vectors in each collectioncollection
qdrant_collection_segmentsnumber of segments in each collectioncollection
qdrant_collection_statuscollection status mapped to a numbercollection · 1=green, 2=yellow, 3=red, 0=unknown

access points

prometheus ports are different

http://localhost:9091 is the host-mapped prometheus ui from docker compose. http://localhost:9090 is the container's internal port — don't use that from your browser.

prometheus scrape config

prometheus inside docker must use the container name, not localhost:

prometheus.yml
scrape_configs:
  - job_name: 'qdrant-exporter'
    static_configs:
      - targets: ['exporter:9999']
    scrape_interval: 15s

grafana dashboard

import the pre-built dashboard in a few clicks:

  1. login to grafana (admin / admin)
  2. go to configuration → data sources → add data source
  3. pick prometheus
  4. set the url to http://prometheus:9090 (internal docker network name)
  5. hit "save & test"
  6. import examples/grafana-dashboard.json
the dashboard will auto-bind to the prometheus data source and render charts for every collection metric.