Usage metrics
Entrepot counts how every repository is used and shows it in the UI — no external monitoring stack needed to answer “which repositories are busy?”, “how much do we serve from cache?” or “who downloaded this?”.
What is counted
Section titled “What is counted”Per repository, Entrepot records:
| Measure | Meaning |
|---|---|
| Requests and errors | every package-client request to the repository (/repository/<name>/…, /v2/<name>/…); errors are server errors (5xx) |
| Downloads and uploads | artifact files served and stored |
| Bytes out / bytes in | traffic served and received |
| Cache hits / misses | proxy lookups answered from the blob store vs fetched upstream — the cache hit ratio |
Counters are aggregated in memory on each replica and flushed every few seconds as additive updates to minute buckets (kept about two days) and hour buckets (kept about 400 days), so any number of replicas can record at once.
The dashboard
Section titled “The dashboard”The UI’s home page (Dashboard) is a usage dashboard over a selectable window (last hour, 24 hours, 7 days or 30 days):
- headline numbers — request rate, requests, downloads, uploads, traffic and cache hit ratio;
- traffic over time;
- the top repositories by activity;
- content inventory per format and components by scan state;
- for administrators, storage per blob store.
What you see depends on who you are: users see totals for the repositories they can read; administrators see every repository, plus storage. Each repository page also has Usage charts for that repository, and an Activity tab.
Activity log
Section titled “Activity log”Uploads and deletes — and, by default, every download — are recorded with
the actor, remote IP, user agent, path and bytes. Administrators see the
whole log (Administration → Activity); a user with admin on a
repository can query that repository. Both the activity log and the audit log
can be filtered and exported as CSV.
# downloads from npm-internal by CI jobs in the last day, as CSVcurl -fsS -H "Authorization: Bearer $ADMIN" \ "https://entrepot.example.com/api/v1/activity?repository=npm-internal&kind=download&actor=gitlab:&from=2026-11-17T00:00:00Z&format=csv"| Endpoint | Returns |
|---|---|
GET /api/v1/metrics/overview?range=24h |
the dashboard data, scoped to readable repositories (scope: readable or all) |
GET /api/v1/repositories/{name}/metrics?range=7d |
one repository’s totals, request rate and time series |
GET /api/v1/activity |
the activity log, filtered by repository, kind, actor, path, from, to; format=csv to export |
range is one of 1h, 24h (default), 7d, 30d.
Configuration
Section titled “Configuration”| Key | Default | Meaning |
|---|---|---|
usage.recordDownloads |
true |
Add every download to the activity log. Counters and charts are kept regardless; turn this off on very busy instances to keep the log small. |
usage.activityRetention |
720h (30 days) |
How long activity events are kept. |
usage.flushInterval |
10s |
How often buffered counters are written to Postgres. |
Prometheus
Section titled “Prometheus”The same counters are exported on /metrics for scraping:
entrepot_repository_requests_total,
entrepot_repository_download_bytes_total,
entrepot_repository_upload_bytes_total and entrepot_proxy_cache_total.
See Monitoring for labels and
queries.