Báo cáo phân tích sau sự cố của Coinbase tiết lộ các rủi ro lỗi lan truyền trên AWS

The Huoxing Finance report states that Coinbase has released a post-mortem report on the large-scale service outage on May 7, 2026. The outage lasted approximately 8 hours, with full recovery taking about 12 hours, during which trading, deposits, withdrawals, and most core services were unavailable or severely degraded. Coinbase stated that the incident was triggered by the simultaneous failure of multiple cooling units in the data center of the availability zone (use1-az4) in the AWS us-east-1 region, causing cabinet thermal protection shutdowns that took down EC2 instances and EBS volumes, subsequently affecting multiple internet services. During recovery, Coinbase’s matching engine lost quorum due to its cluster architecture being deployed in a single AWS data center, requiring emergency code adjustments and the rebuilding of a new node group to restore operations, with market trading gradually restarted during the recovery process. Additionally, AWS-managed Kafka (MSK) experienced a control plane failure, preventing automatic re-election of partition leaders, further blocking quote feeds, fee calculations, and certain settlement and data flow systems, expanding the overall impact. After collaborative manual partition migration by Coinbase and AWS engineering teams, the system gradually returned to normal. Coinbase acknowledged that this incident exposed deficiencies in its cross-availability zone automatic failover capabilities and disaster resilience of managed middleware. The company will upgrade its cross-region hot-standby architecture, strengthen regular failure drills, migrate its Kafka system from a dual-availability zone to a tri-availability zone deployment, and work with AWS to address root causes and implement improvements.