Ntfy
Self-hosted push notification service. Ntfy receives HTTP POST messages and delivers them to subscribed clients (mobile apps, web UI, CLI).
Quick Reference
| Property | Value |
|---|---|
| URL | https://ntfy.ops.eblu.me |
| Tailscale URL | https://ntfy.tail8d86e.ts.net |
| Namespace | ntfy |
| Image | registry.ops.eblu.me/blumeops/ntfy:v2.28.0-2288d19-nix (locally built) |
| Upstream | https://github.com/binwiederhier/ntfy |
| Manifests | argocd/manifests/ntfy/ |
| Topic declaration | ntfy/topics.json |
Architecture
Ntfy runs as a single pod with no persistent storage — message cache and attachments use an emptyDir volume. This is intentional: ntfy is treated as an ephemeral delivery channel, not a message store. Messages lost on pod restart are acceptable.
The upstream relay (ntfy.sh) is configured so mobile app clients can receive push notifications via Google FCM / Apple APNs without self-hosting those integrations.
Topics
Mirrors ntfy/topics.json by convention — the declared-state file that is the single source of truth for topics; the declaration itself is enforced by mise run ntfy-lint (an undeclared or retired topic literal in the tree fails CI).
| Topic | Purpose | Volume | Priority | Actionable | Producers |
|---|---|---|---|---|---|
ops-approvals | Warrant requests, auto-apply refusal/failure alerts, the weekly auto-apply digest, host apply verdicts (ringtail / indri rebuild watchers), forge push-mirror guard alerts | a handful per day; bursts on rebuild failures | 3 (host watchers + mirror guard 4) | yes | .forgejo/workflows/auto-apply-digest.yaml, .forgejo/workflows/provision-indri.yaml, .forgejo/workflows/ringtail-rebuild.yaml, mise-tasks/request-run, the two host apply watchers, darwin/indri/mcquack-forgejo-mirror-guard.sh |
infra-alerts | Grafana Unified Alerting pipeline: service health, pod readiness, PostgreSQL, textfile freshness, Frigate cameras, ArgoCD sync status | a few per day; group_wait 1m, group_interval 12h, repeat_interval 24h | 3 | yes | argocd/manifests/grafana-ringtail/alerting.yaml |
frigate-alerts | Frigate camera detection events (person, car) in the driveway zones, with action buttons into the review UI | a few per day in driveway zones only | 3 | no | argocd/manifests/frigate/frigate-notify-config.yml |
ceramics | Skagit CCE ceramics watch — retired 2026-09-01, kept declared so a silent re-addition fails the lint | none — producer retired | 3 | (was yes) | none |
Adding a producer to a topic
A recurring notification on a phone-subscribed topic is a reviewed change to ntfy/topics.json: update the topic’s producers and budget.volume there (and this table), then add the literal in the code. An undeclared topic literal fails mise run ntfy-lint (prek hook + Ntfy Lint CI). The budget columns exist so notification creep is visible in the diff: if the cadence grows, the declaration grows with it.
Subscribe to
The one list a new phone or device setup needs. Add each topic in the ntfy app (server: https://ntfy.ops.eblu.me):
-
ops-approvals— warrant requests, auto-apply alerts, host apply verdicts. Actionable: sound + banner always on; this is the topic that carries the approval gate. -
infra-alerts— Grafana alerting pipeline with runbook links. Actionable, but repeats are the norm (24h repeat interval). -
frigate-alerts— driveway camera detections. FYI; consider a distinct sound or lower volume. -
ceramics
Producers
Every producer posts to a declared topic. The topic literals live in the producers themselves (shell watchers, workflows, the request-run task, and the argocd/ manifests that lint-only by design — ArgoCD owns that tree); ntfy/topics.json declares which topics exist, what posts to them, and at what budget.
- Host apply watchers (
nixos/ringtail/ringtail-apply-watcher.sh,darwin/indri/mcquack-indri-rebuild-watcher.sh) post apply verdicts;NTFY_TOPICenv override, default the declared topic. mise-tasks/request-runposts a notification per--notify.- CI workflows (
auto-apply-digest.yaml,ringtail-rebuild.yaml,provision-indri.yaml) post the digest and refusal/failure alerts. - Grafana posts via the webhook contact point in
argocd/manifests/grafana-ringtail/alerting.yaml(topic set in the message template). - frigate-notify posts detection alerts via
argocd/manifests/frigate/frigate-notify-config.yml(cluster-internal addresshttp://ntfy.ntfy.svc.cluster.local:80).
Configuration
Server config is in a ConfigMap (ntfy-config):
| Setting | Value |
|---|---|
base-url | https://ntfy.ops.eblu.me |
upstream-base-url | https://ntfy.sh |
attachment-total-size-limit | 1 GB |
attachment-file-size-limit | 10 MB |
attachment-expiry-duration | 24h |
No authentication is configured — access is restricted by Tailscale ACLs (only tailnet clients can reach the service).
Related
- routing - How ntfy is exposed via Caddy
- observability - Monitoring and alerting infrastructure
- declared-state-files - The convention
ntfy/topics.jsonfollows