- Sources: Google Cloud status incident, HN discussion
- Summary: The status page records the incident as running from 08:40 to 12:20 US/Pacific on 2026-08-20, with mitigation stated as of 10:22 PDT, so it is the previous day rather than the date of this digest. Google names the cause in the status update rather than deferring it. A planned optical maintenance removed capacity and caused unexpected congestion in the Dalles, Oregon metro, and services recovered once capacity was restored. The listed blast radius covers GKE, Cloud Run, Cloud SQL, AlloyDB, BigQuery, Cloud Storage, Pub/Sub, Compute Engine, Persistent Disk and IAM. Google's own workaround during the incident was to fail over to another region. A full analysis is promised and not yet published.
- Why it matters: Planned network maintenance took 27 products down in one region, so anyone single-homed in us-west1 had no path that did not cross the congested metro.
- Follow-up: Read the full incident analysis when Google publishes it.
send feedback on this story