Join on WhatsApp
Join on Telegram
Mohammedia – Amazon Web Services (AWS) has shared new details about what caused the major service outage that hit its Northern Virginia (us-east-1) region on October 19 and 20, 2025. The problem started late at night and lasted more than 14 hours, affecting some of AWS’s biggest services, including DynamoDB, EC2, and Lambda, and disrupting many customer applications around the world.
The issue began at 11:48 PM PDT on October 19, when DynamoDB, one of AWS’s main database services, started returning high error rates.
AWS said the outage happened because of a software bug, or “race condition,” in its automated DNS system—the part of the service that connects user requests to servers.
This bug caused DynamoDB’s main regional address (dynamodb.us-east-1.amazonaws.com) to lose all its IP addresses, meaning computers could no longer find or connect to the database.
Outage spreads across core AWS services
Since many AWS services depend on DynamoDB, the problem quickly spread. EC2, the service that runs virtual servers, started failing to launch new instances because its internal management system also relies on DynamoDB.
While servers that were already running stayed online, customers couldn’t create new ones for several hours. The recovery of EC2 wasn’t complete until 1:50 PM the next day.
The outage also affected AWS’s Network Load Balancer (NLB), which helps direct traffic between servers. From early morning until early afternoon on October 20, some NLBs had trouble checking the health of new EC2 instances.
Because of delayed network updates, the system mistakenly marked healthy instances as broken and removed them from service. This caused connection errors and performance drops for several applications that use NLB.
Other AWS services also went down or slowed significantly. Lambda, which runs serverless code, had trouble processing requests.
Container services like ECS, EKS, and Fargate experienced failures and delays. Even Amazon Connect, AWS’s customer service platform, saw call and chat disruptions that left some users unable to reach support or sign in to their accounts.
AWS engineers discovered the main cause of the issue about an hour after it began. They had to manually fix DynamoDB’s DNS records to restore connections.
By 2:25 AM on October 20, the database was back online, and most other services started to recover over the next several hours.
To prevent this from happening again, AWS has turned off the part of DynamoDB’s system that automatically updates DNS records until it is fixed.
The company said it will patch the software bug and add extra protections so outdated or incorrect DNS updates can’t be applied. It will also add new safety checks for EC2 and better controls for NLB to limit the impact of future failures.
AWS apologized to customers for the disruption, saying it understands how much businesses depend on its services. “We know this event impacted many customers in significant ways,” the company said. “We’re learning from this and will continue improving to make our systems even more reliable.”
