Teleport Outage History

A regression introduced in Teleport v18.7.6 broke the Just-in-Time \(JIT\) access flow when connecting to database agents running versions earlier than 18.7.6. The defect affected clusters on versions 18.7.6, 18.8.0, and 18.8.1. The 18.8.0 and \(upcoming\) 18.8.1 tenant upgrades would have impacted ~16% of Cloud customers, specifically those using database access. To prevent disruption, affected customers were rolled back to a stable version while the fix lands in 18.8.2.

On January 6, 2026 at 18:24 UTC, approximately 24% of Teleport Cloud customers experienced an interruption to Teleport Auth and/or Proxy services for up to 24 minutes. The incident was caused by two concurrent infrastructure events that compounded each other. A resource exhaustion issue on multiple nodes temporarily disrupted scheduling. At the same time, an unrelated scaling event interrupted a backend service responsible for refreshing cloud credentials used by Teleport Auth, which led to authentication failures. Once the affected infrastructure was removed, service recovery began immediately. Backend services were restored, credentials were refreshed, and impacted Teleport services gradually returned to a healthy state. We are making changes to improve isolation between critical services, enforce stricter resource limits, and ensure that a single infrastructure disruption cannot prevent credential refresh or delay service recovery in the future.

On October 20, 2025, between 06:48 and 22:07 UTC, some Teleport Cloud customers experienced issues connecting to their cluster. These issues were triggered by an AWS outage impacting the US East region \(N. Virginia: us-east-1\). The severity of the impact was dependent on the type of Teleport Cloud deployment and where users or agents were located. * Customers with single-region auth service deployments in the US East region were unable to connect to their cluster * Customers with multi-region auth service deployments in the US East region experienced intermittent connectivity issues * Users and agents located near the US East region experienced intermittent connectivity issues For more details, please download the full RCA: [RCA: 2025-10-20 US East Region Connectivity](https://trust.goteleport.com/resources?s=ku9gorp0blta6u96pazrhw&name=teleport-cloud-rca:-2025-10-20-us-east-region-connectivity)

On October 20, 2025, between 06:48 and 22:07 UTC, some Teleport Cloud customers experienced issues connecting to their cluster. These issues were triggered by an AWS outage impacting the US East region \(N. Virginia: us-east-1\). The severity of the impact was dependent on the type of Teleport Cloud deployment and where users or agents were located. * Customers with single-region auth service deployments in the US East region were unable to connect to their cluster * Customers with multi-region auth service deployments in the US East region experienced intermittent connectivity issues * Users and agents located near the US East region experienced intermittent connectivity issues For more details, please download the full RCA: [RCA: 2025-10-20 US East Region Connectivity](https://trust.goteleport.com/resources?s=ku9gorp0blta6u96pazrhw&name=teleport-cloud-rca:-2025-10-20-us-east-region-connectivity)

Teleport Cloud customers experienced a service disruption intermittently impacting connectivity for about 40 minutes. The root cause of the incident was a bug in the Envoy Gateway ingress service that resulted in excessive memory utilization and ultimately restarts. The incident was triggered by the restart of all tenant services due to an undocumented background EKS storage migration that occurred during a scheduled maintenance period to upgrade the Kubernetes control plane. In response to the incident, the Teleport team shut down ingress controllers which prevented routing updates from triggering ingress service OOM events. This reduced customer impact by stabilizing ingress connectivity. Once the EKS storage migration and tenant service restarts were completed, the team rolled back the Envoy gateway version which stabilized the platform.

A small number of customers \(<5\) with a custom backend configuration experienced a connectivity issue in us-west-2 lasting around 30 minutes due to a Cilium failure that was inadvertently blocking connections between Auth servers and their backend.