We recently addressed issues affecting the Box API, All Files, and Box AI. We would like to take the opportunity to further explain what happened and the steps we have taken to prevent reoccurrence.

Between 12:06 PM PDT and 12:39 PM PDT on June 17, 2026, some users may have experienced difficulties while working in Box. During this time, some customers were unable to access content, browse folders, or complete uploads and other file operations, and error rates for affected features were elevated.

What Happened

The issue occurred after a recent rollout of a new traffic-routing change for the All Files experience reached full traffic volume. This routing change directed file-browsing requests through a newer service path that, under full production load, generated a significantly higher volume of calls to an underlying data-access component than had been observed during earlier, partial rollouts. The increased call volume saturated a capacity-limited thread pool within our file service. Once that threshold was crossed, automated health checks detected the degraded state and began restarting affected service instances — which, combined with retry traffic from dependent services, amplified the load and caused a cascading degradation affecting the API, All Files browsing, and Box AI.

How We Restored Service

We restored service by:

• Disabling the recent traffic-routing rollout, returning file-browsing requests to the prior service path
• Increasing the capacity of the affected file service to absorb the load
• Reverting a separate, unrelated deployment as a precautionary measure

Monitoring confirmed that availability and error rates returned to normal by 12:39 PM PDT, approximately 33 minutes after impact began.

Analysis

This incident highlighted opportunities to evaluate how a new traffic-routing change is validated at production scale and how dependent components behave when exposed to that traffic pattern for the first time. Specifically, the new routing path generated a more resource-intensive class of data queries than the partial rollout had revealed, and the affected service lacked sufficient capacity headroom to absorb the full traffic volume. It also showed the value of rapid rollback capability and coordinated monitoring across feature rollouts and service capacity limits.

Corrective Actions

Box has initiated the following corrective actions:

• Increase and harden service capacity and autoscaling: We have increased the capacity of the affected file service and are improving autoscaling thresholds so the service can respond dynamically to traffic spikes.
• Improve pre-rollout validation: We are strengthening load testing and validation of traffic-routing changes under realistic, production-like conditions before full deployment.
• Tighten rollout safeguards and monitoring: We are improving rollout gates, alerting, and automated rollback capabilities to enable faster detection and response when changes cause elevated error rates.
• Decouple health probes from capacity saturation: We are updating how service health checks interact with capacity limits to avoid crash-loop scenarios that can amplify failures during high-load events.

We are continuously working to improve Box and want to make sure we are delivering the best product and user experience we can. We hope this provides clarity on what occurred and the steps we are taking. We would be happy to answer any questions you may still have regarding this matter.

Sincerely, The Box Team