Nostrfi Relay 0.2.0: run your own Nostr relay and see what it is doing
· Nostrfi
Nostrfi Relay is a self-operable Nostr relay with an operator dashboard. Version 0.2.0 is released, it is open source, and Nostrfi runs it in production. Anyone can run it today.
Why we built it
Starting a Nostr relay is easy. Operating one well is harder. An operator has to get protocol compatibility, storage, moderation, rate limits, monitoring, backups and capacity right, often without much visibility into what the relay is actually doing. Who is connected? What is being stored? What was rejected, and why? How much headroom is left? Will the backup restore?
Nostrfi Relay is built to answer those questions. It pairs a compatible relay with a dashboard for the person who runs it.
Who it is for
- Client and app teams that need a relay they control, for availability, data policy or interoperability testing.
- Communities and organisations that want their own relay with their own moderation and retention rules.
- Individual operators running public relay infrastructure.
If you are looking for a Nostr client or a hosted account, this is not it.
What you get
- A relay that works with compatible Nostr clients. It supports NIP-01, 02, 09, 11, 17, 22, 28, 40, 42, 50, 70, 71, 77, 86 and 98, including NIP-42 authentication, NIP-50 search and NIP-77 Negentropy sync. The protocol support page states the limits and what is still being tested.
- An operator dashboard you sign in to with your own Nostr key, through a NIP-07 or NIP-46 signer. It shows live connections, rates and rejections, stored events by period and kind, and the configuration the relay is actually enforcing. See the admin dashboard.
- An event browser for finding stored events by kind, author, content, time range and tags. See the event browser.
- Moderation through NIP-86. Ban pubkeys and events, and block IPs, from the dashboard or any NIP-86 client. See moderation.
- Backups you can check. A restore is checked against row counts, not assumed. See backup and retention.
- Measured capacity. The capacity page publishes memory and throughput figures from load tests in a development sandbox, and how to reproduce them on your own hardware.
What is new in 0.2.0
- The event browser filters by several tags at once, including multi-letter tags.
- The Overview and Metrics pages show relay status and uptime at a glance.
- The relay answers every NIP-42
AUTHmessage with anOK, so clients that wait for it, such asnakand go-nostr, no longer time out.
The changelog has the details for operators.
What it does not do yet
We publish the limits as well as the features. The main ones today:
- It runs as a single instance, by design. Nostr gets its redundancy from clients using several relays, so most operators never need to scale one out.
- Search results are not ranked.
- There is no encryption at rest and no TLS in the application. Put it behind a reverse proxy, as the quickstart does.
- There is no service-level objective yet.
The full list, with the reasons and any compensating controls, is on the current limitations page. The trust and data page explains what a relay operator can see, and links to the threat model.
Run it
The quickstart takes you from nothing to a running relay with its dashboard and TLS, using Docker Compose, in about ten minutes. You then sign in with your Nostr key and connect a client.
The source is on GitHub. Issues and questions are welcome there.