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

Loss of network connectivity in US West (Oregon)

Instances in US West (Oregon) were unreachable for 120 minutes. No instance restarted.

Reference 3c929029f6e6 · US West (Oregon) · Updated just now

Status
ResolvedResolved — last update 28 Aug 2026, 15:49 UTC
Started
28 Aug, 13:49Friday 28 August 2026 UTC
Duration
2 hEnded 28 Aug 2026, 15:49 UTC
Components
3Network and public IPs — US West (Oregon), Instance launch and lifecycle — US West (Oregon), Spot pool and reclaim notices — US West (Oregon)
Timeline

What we posted, as we posted it.

Resolved4 updates · UTC
  1. ResolvedThis incident has been resolved. Affected accounts have been credited for the unreachable minutes. A post-incident review is published below.28 Aug, 15:49
  2. MonitoringConnectivity has been restored since 15:02 UTC. Instances did not restart and disks were not affected. We are monitoring the pair.28 Aug, 15:02
  3. IdentifiedThe core switch pair in US West (Oregon) failed over and the standby came up with an incomplete configuration. Connectivity is being restored from the last known good configuration.28 Aug, 14:29
  4. InvestigatingInstances in US West (Oregon) are unreachable from the Internet. Instances are still running; this is a network event. We are working on it.28 Aug, 13:49

Post-incident review

published 2 Sep 2026, 00:49 UTC
Impact
Instances in US West (Oregon) were unreachable for 120 minutes. No instance restarted.
Root cause
Configuration synchronisation between the two core switches had silently stopped nine days earlier.
What we changed
Configuration sync is now checked every five minutes, a mismatch pages the on-call engineer, and failover is exercised monthly.
Credits
Accounts with affected instances were credited for the interrupted minutes under the SLA, without having to ask.

Details

Type
Major outage
Started
28 Aug 2026, 13:49 UTC
Ended
28 Aug 2026, 15:49 UTC
Region
US West (Oregon)
Impact
Instances in US West (Oregon) were unreachable for 120 minutes. No instance restarted.
Reference
3c929029f6e6

Follow the Atom feed