System Status
Real-time health of the data plane: recorders, caches and streams. Snapshot auto-refreshes every 60 seconds via SSR revalidate.
Components
Per-component health: thresholds are conservative. "Degraded" means slower than expected, "offline" means stale beyond the degraded ceiling.
Data watchdog (hourly)
Our own tables checked against their expected cadence every hour: a venue's open interest jumping more than 30% between snapshot buckets, a counted liquidation feed going silent, a stale board cache, a late worker job, a missing daily metric. Findings stay here until they clear.
Last check 2026-10-23 09:51 UTC
| Level | Check | Subject | Detail |
|---|---|---|---|
| warn | feed_stale:orderbook_depth | weex | no record in the registry's lookback window |
How this is measured
Heartbeats come from the internal worker_state table updated by the worker, the streamer and the market recorders. A component is degraded when its last update is older than the healthy ceiling, and offline when older than the degraded ceiling. DB latency is a live SELECT 1 measured each render. Open incidents are read from the transparency log. See methodology.
This page reports platform health, not market conditions. A green status does not constitute investment advice.