Coinbase: Configuration Error Knocked Out Gateway for 50 Minutes

Coinbase traced a roughly 50-minute outage to a Kubernetes resource name collision that occurred during a routine configuration update and misconfigured its Istio ingress gateway. The incident prevented internal systems from communicating and affected transfers, Coinbase Card debit purchases and onchain services from about 12:37 p.m. ET to 1:25 p.m. ET.

Engineers deploying what the company described as a low-risk update accidentally targeted the wrong network component after two Kubernetes resources shared the same name. The update altered part of the Istio ingress gateway, a traffic controller that routes requests between internal services, which knocked the gateway offline and blocked systems that process trades, transfers, settlements and payment authorizations.

The outage disrupted retail, institutional and developer platforms. Some retail customers could not complete deposits, withdrawals or off-platform transfers; transfers already underway appeared stuck but were processed after services returned. Coinbase Card debit transactions were declined while credit card purchases continued to process. Onchain swaps through Coinbase DEX on Base and Solana were interrupted. Exchange and Prime customers experienced delayed or failed transfers and settlements. Developer-platform users could not onboard customers, move funds, or use onramp services during the outage.

Coinbase restored the gateway at 1:20 p.m. ET and declared the incident mitigated at 1:23 p.m. ET. Remaining transaction backlogs cleared over the following hours. The postmortem noted that customer funds were not at risk.

Recovery was slower because the company’s standard deployment and rollback tools relied on the affected gateway and could not be used. Engineers initiated an emergency deployment through the cloud provider using temporary privileged access to restore the gateway.

The postmortem outlined steps to reduce the chance of a repeat: add checks to prevent naming conflicts in Kubernetes, separate recovery tools from the infrastructure they manage, and increase regular testing of emergency access procedures. The company wrote, “This was caused by an unintended misconfiguration of an important network component as a result of a routine configuration update,” and added, “Our goal is always zero downtime. We are actively redesigning our infrastructure to ensure that, in the highly unlikely event of another failure like this one, we can swiftly revert and recover.”

Kubernetes is the software Coinbase uses to run applications across cloud servers; a resource name collision happens when two resources share the same identifier. Istio is a service mesh that manages traffic between services; its ingress gateway directs incoming requests to the correct internal components. In this incident, the update targeted the ingress gateway rather than the intended component, which cut off internal service communication.