admin · system
Drift-status
Kerne-services: oceaneye-backend (CPX32), tile-værksted (CX22), R2 storage, AIS-pipeline. Auto-refresh hvert 30s.
tiles.oceaneye.blue
checking
Endpointtiles.oceaneye.blue
BackendCloudflare R2 + edge
Buckets—
Sidste check—
api.oceaneye.blue
checking
Endpointapi.oceaneye.blue
BackendHetzner CPX32 (46.225.181.103)
StackPostgres + Valkey + workers
Sidste check—
AIS-pipeline (live)
Positions / 5 min
—
via ingress + ingest-api
Positions / 1 t
—
over alle sources
Active vessels / 1 t
—
unique MMSIs i boats
Open alerts
—
ntfy via ingress-monitor
Sidste opdatering: — · auto-refresh 30s
Engine room (CPX32)
Worker pending
—
Valkey XPENDING — er DB ved at hænge?
Stream length
—
stream:positions buffer
Valkey memory
—
used_memory
Postgres DB
—
pg_database_size
Worker pending > 50k = ERR; > 10k = WARN. Stream length sidder naturligt på MAXLEN 1M (rolling buffer, ikke backlog).
Smart pipeline (Stage 0 fysik + Stage 2 geofence)
Noise / 5 min
—
drop: speed>100kn, invalid lat/lon
Suspect / 5 min
—
flag: on_land via geofence
Critical unreviewed
—
manuel review pending
Total / 24 t
—
cron sletter noise>30d, suspect>90d
Noise = AIS-fix dropped (gemmes ikke). Suspect = beholdt men flagged (typisk on_land suspects ~30-40/min). Critical > 0 = manuel handling.
Boats-coverage (type 5 / 24 metadata)
Boats total
—
unique MMSIs nogensinde set
Med navn
—
decoder leverede ship_name
Coverage
—
% af boats med rigtigt navn
Placeholder
—
"MMSI xxx" — venter på type 5
Alerts (sidste 24t)
Henter alert-historik …
AIS-sources
Henter sources …
Klik en source for detaljeret graf og reject-grunde sidste time.
Tile-pipeline (auto-heal)
Henter status …
Storage
Henter manifest …
Hosts
| Host | IP | Rolle | Specs |
|---|
| oceaneye-prod |
46.225.181.103 |
Hot-path: API, DB, workers, Centrifugo |
CPX32 · 4 vCPU · 8 GB · 160 GB NVMe |
| aiss-aisstream |
178.104.250.139 |
AIS collector + tile-build værksted |
CX22 · 2 vCPU · 4 GB · 38 GB |
| R2 oceaneye-tiles |
Cloudflare EEUR |
Public PMTiles serving |
1.5 GB used · gratis egress |
| R2 oceaneye-backups |
Cloudflare EEUR |
pg_dump + Valkey RDB cold storage |
Vokser · gratis egress |
Cron timers (Hetzner CX22)
| Timer | Schedule | Funktion |
|---|
oceaneye-auto-heal.timer | 06:00 UTC daglig | Tile-freshness-check + auto-rebuild stale |
Cron timers på oceaneye-prod CPX32 (DB-backups, Valkey RDB, monitor) ses ikke her — tjek oceaneye status CLI eller SSH til boxen.