Cloudflare Outage History

Cloudflare experienced elevated error rates in its Workers AI service affecting customers using the Llama 4 Scout model (model @cf/meta/llama-4-scout-17b-16e-instruct). The issue began at Jul 10, 2025 7:32 PM UTC and Cloudflare flagged it as a warning at Jul 10, 2025 7:35 PM UTC as investigations started. The problem impacted the Cloudflare Sites and Services / Workers AI component, causing failed or slower model invocations for a period of about 1 hour and 13 minutes, with restoration by Jul 10, 2025 8:45 PM UTC. Cloudflare issued two status updates during the incident to inform users of the investigation and restoration progress.

The Cloudflare Durable Objects service experienced an incident affecting the Asia-Pacific region beginning at 2:45 PM UTC. Users reported increased errors and degraded performance for workloads relying on Durable Objects, with activity centered in the Asia-Pacific region and Mumbai.

Cloudflare began a scheduled maintenance window in the Halifax region (YHZ) at Jul 9, 2025 5:00 AM UTC. The status page noted: 'In progress - Scheduled maintenance is currently in progress. We will provide updates as necessary.' The maintenance was localized to the Halifax area and could affect services routed through this region.

The maintenance concluded with the region restored by Jul 9, 2025 2:05 PM UTC, for a total duration of about 9 hours and 5 minutes.

Cloudflare experienced degraded performance affecting zone plan updates during scheduled maintenance in the New Delhi region. Some customers attempting to set or change zone plans encountered HTTP 500 errors, impacting Zone Plan management within Cloudflare Sites and Services / Subscriptions. The outage began at 8:30 PM UTC and was resolved by 4:00 AM UTC.

Cloudflare issued four status updates during the incident, including notes on the New Delhi maintenance and the Zone Plan Update Errors, and the affected component Cloudflare Sites and Services / Subscriptions. The issue appeared regional, with impact primarily in the DEL region, and service returned to normal after maintenance concluded.

Cloudflare conducted scheduled maintenance in the STI region (Santiago de los Caballeros). The status page indicated the work was in progress and that services could experience temporary degradation while the maintenance was under way. The maintenance began around 6:01 AM UTC and the status shifted to degraded at 6:05 AM UTC. The maintenance window lasted roughly 5 hours, with services returning to normal by 11:05 AM UTC. The incident was localized to the STI region, and Cloudflare posted two official updates during the period to communicate progress and completion.

Cloudflare carried out scheduled maintenance in the New Delhi region (DEL), which caused a brief degradation for users in that area. The New Delhi region entered a degraded state at Jul 7, 2025 8:35 PM UTC as maintenance began, with the status page noting that maintenance was in progress and updates would be provided. The maintenance continued for about 6 hours and 40 minutes, and the Delhi region returned to normal after Jul 8, 2025 3:15 AM UTC.

Cloudflare conducted scheduled maintenance in the YYZ (Toronto) region, which led to degraded service for users in that area. The incident began at 8:05 AM UTC when the status page updated to indicate maintenance in progress, and it persisted for about 3 hours and 55 minutes until it was resolved at 12:00 PM UTC. The change appeared to be region-specific to YYZ.

Cloudflare posted two updates during the incident, confirming that scheduled maintenance was in progress and later noting that the service had returned to normal. No user reports were observed during this event, and the impact appeared limited to the YYZ region.

Cloudflare performed scheduled maintenance at its Baghdad gateway (BGW) starting at 3:00 AM UTC. The operation briefly affected traffic routed through the Baghdad edge location and lasted 2 hours 35 minutes, with services restored by 5:35 AM UTC. Cloudflare issued 2 status updates during the incident to communicate in-progress work and completion. The impact appeared localized to the Baghdad gateway region, while global Cloudflare services remained operational.

An extended Cloudflare incident began at Jul 2, 2025 3:00 AM UTC and affected a broad set of services across many regions, including Bring Your Own IP (BYOIP), R2, WARP, Durable Objects, Firewall, and general network performance. Users worldwide experienced degraded performance and intermittent connectivity as Cloudflare components used by apps and services encountered issues. Regions such as North America, Europe, Asia, and others reported impact, indicating a broad global footprint. The outage lasted roughly 2 days and 19 hours, with recovery milestones noted as updates progressed toward normal operation and the final recovery confirmed by Jul 4, 2025 9:50 PM UTC.

Cloudflare provided regular status updates throughout the incident, detailing investigations, mitigations, and in some cases configuration rollbacks to address the disruption. The event involved multiple components and regions and prompted ongoing communications until most services were restored.

User reports during the incident were limited in volume, but some users did report connectivity problems and degraded performance consistent with the documented issues.

Cloudflare’s Durable Objects in the WEUR region experienced a minor availability impact, with some operations potentially unreliable during the incident. The outage began around 5:40 PM UTC and was resolved by 8:15 PM UTC.

Cloudflare issued 2 status updates during the incident. The first update at Jul 1, 2025 5:45 PM UTC noted an investigation into the minor impact, and the service returned to normal by Jul 1, 2025 8:15 PM UTC.