Coinbase has published a post-mortem explaining the 50-minute service disruption that hit its platform on July 14, affecting retail and institutional trading, deposits, withdrawals, card transactions, and onchain swaps. The outage, which lasted from 12:37 PM to about 1:25 PM ET, was traced to a routine configuration update that inadvertently caused a resource name collision in a production Kubernetes cluster.
According to Coinbase, the change was part of a planned migration to a new service deployment model and was considered low-risk. However, a conflict in the Istio ingress gateway went undetected during pre-production checks, blocking internal network access and halting nearly all asynchronous workflows. These workflows handle everything from trade settlement to card authorizations. "When infrastructure services became unreachable, those workflows paused platform-wide," the company stated.
User funds were never at risk, and any in-flight transactions completed once service was restored. The recovery was delayed because the same gateway also managed deployment tools, forcing engineers to use emergency procedures. In response, Coinbase announced several infrastructure upgrades: stronger deployment safeguards to detect Kubernetes resource conflicts, greater separation between deployment systems and managed infrastructure, and regular testing of emergency recovery mechanisms. The company reiterated its goal of "zero downtime" and apologized for the inconvenience.
The incident, while quickly resolved, underscores the operational complexity of running a platform of Coinbase's scale. The exchange has been expanding into stock trading and institutional services, and it follows two consecutive outages on its incubated Base blockchain in late June. Shares of COIN were up over 11% on the day of the report.
