Skip to content
OutageDeck

Incident detail

Emerging Service Issue - GPU Deployment

Emerging Service Issue - GPU Deployment

Resolved incidentMinor0 affected services

Timeline window

to

Timeline

Incident updates

Updates are normalized from the official source chronology so timeline changes remain easy to scan.

  1. Identified

    Our team is investigating an emerging service issue affecting GPU provisioning in all regions. At this time deployment of RTX PRO 6000 Blackwell plans in unavailable. We will share additional updates as we have more information.

  2. Identified

    The issue with deployment of RTX PRO 6000 Blackwell plans appears to have started around 15:00 UTC on May 19, 2026. We are actively working to deploy an API hotfix that is expected to be rolled out within 1 hour. We apologize for the impact.

  3. Monitoring

    At this time we have been able to correct the issues affecting GPU provisioning in all regions as of 21:27 UTC. on May 19th, 2026. We will be monitoring this to ensure that it remains stable. If you continue to experience problems, please <a href="https://cloud.linode.com/support/tickets">open a Support ticket</a> for assistance.

  4. Resolved

    We haven’t observed any additional issues with the GPU provisioning in all regions, and will now consider this incident resolved. If you continue to experience problems, please <a href="https://cloud.linode.com/support/tickets">open a Support ticket</a> for assistance.

  5. Resolved

    On May 19, 2026, at 18:26 UTC, we identified an issue impacting the creation and re-provisioning of Linode instances using any of the four RTX PRO 6000 Blackwell \(g3-gpu-rtxpro6000-blackwell\) plans, resulting in API errors. These failures caused a denial of service for customers attempting to deploy g3-gpu-rtxpro6000-blackwell instances via the Cloud Manager or API. After the initial investigation SMEs determined that the cause of the issue was a logic error within a recent APIv4 update. Specifically, a validation check intended only for RDMA-enabled networking was incorrectly applied to standard Blackwell GPU plans. This caused the API to reject creation requests that did not meet specific RDMA configurations, even when those configurations were not required for the selected plan. With the issue identified, our engineering team developed and tested a hotfix to correct the plan-validation logic. Deployment of the fix began shortly after, and broader customer impact was mitigated as of 21:36 UTC on May 19, 2026, following successful verification in all regions. This summary provides an overview of our current understanding of the incident given the information available. Our investigation is ongoing, and we are currently reviewing our automated testing to prevent similar recurrences.