New B200 spot capacity is live in US East from $1.69 per GPU-hour. See availability
Degraded performance

Increased API latency

API p95 latency was 3× normal for 42 minutes; no request failed.

Reference 598cce938db8 · Updated just now

Status
ResolvedResolved — last update 4 May 2025, 22:58 UTC
Started
4 May, 22:16Sunday 4 May 2025 UTC
Duration
42 minEnded 4 May 2025, 22:58 UTC
Components
1API
Timeline

What we posted, as we posted it.

Resolved4 updates · UTC
  1. ResolvedThis incident has been resolved.4 May, 22:58
  2. MonitoringLatency is back within normal bounds since 22:49 UTC. We are monitoring replica lag before moving reads back.4 May, 22:49
  3. IdentifiedThe read replica used for catalogue and instance listings fell behind after a burst of writes and started serving from disk. Reads have been moved to the primary.4 May, 22:28
  4. InvestigatingAPI response times are elevated: p95 latency is around 3× its usual value. Requests are succeeding, only slower.4 May, 22:16

Affected components

Details

Type
Degraded performance
Started
4 May 2025, 22:16 UTC
Ended
4 May 2025, 22:58 UTC
Impact
API p95 latency was 3× normal for 42 minutes; no request failed.
Reference
598cce938db8

Follow the Atom feed