Stats Dashboard
Features

Stats Dashboard

Live pull metrics, top repositories, top clients, cache health, and HTTP status distribution.

Using an AI assistant? Give it this file.

/llms.md is a single Markdown file covering installation, configuration, quickstart, clients, upstreams, permissions, TLS, and production deployment. Copy it and paste it into your AI chat, or tell the AI to fetch https://drevnix.tech/llms.md.

Stats Dashboard

The Stats Dashboard provides a live view of proxy activity and system health. Use it to monitor pull volume, identify heavily-used images, spot misbehaving clients, and track cache effectiveness.

Access the dashboard from Stats in the portal sidebar. The stats:view permission is required.

Pull metrics

The top section shows pull volume over time:

  • Total requests: all proxy requests in the selected time window
  • Successful pulls: requests that returned 200
  • Denied requests: 401 (auth failed) and 403 (permission denied)
  • Cache hit rate: percentage of blob requests served from the local cache (only shown when cache is enabled)

Use the time range selector to view the last hour, 24 hours, 7 days, or 30 days.

Top repositories

A ranked list of the most frequently pulled repositories in the selected window, showing the pull count and which upstream each repository belongs to.

Use this to:

  • Identify which images are candidates for caching
  • Spot unexpected access to sensitive repositories

Top clients

A ranked list of the clients generating the most pull requests. Use this to:

  • Identify unusually high-volume clients (potential misconfiguration or runaway pipeline)
  • Understand which teams or pipelines are the heaviest consumers

Cache health

When pull-through caching is enabled, the dashboard shows:

  • Disk usage: current cache size vs. configured maximum
  • Cached entries: total number of blobs in the cache
  • In-progress writes: blobs currently being written (a non-zero value means pulls are actively populating the cache)

If disk usage is near the maximum, consider increasing CACHE_MAX_SIZE_BYTES or clearing infrequently-accessed entries from the Cache management page.

HTTP status distribution

A breakdown of all proxy responses by HTTP status code:

StatusMeaning
200Successful pull
401Authentication failed (bad credentials or expired client)
403Permission denied (no matching permission for the repository)
404Upstream ID not found in the pull URL
5xxServer-side errors (upstream registry unreachable, database issue, etc.)

A spike in 401 or 403 responses is worth investigating. It often means a CI pipeline has stale credentials or a recently deleted client is still trying to pull.