Amazon Web Services experienced an operational incident in its US-EAST-1 region in Northern Virginia on Sunday, 15 November 2026. The AWS US-EAST-1 outage blocked sign-ins to the AWS console, stopped new account sign-ups and broke network connections across 11 services. Authentication stayed down for 2 hours 15 minutes with no workaround at all, and a residual network-connectivity problem continued for another hour after sign-ins recovered. By 6:53 AM Pacific time, AWS said every service was operating normally again.
The AWS US-EAST-1 outage timeline: two phases, one incident
The outage began at 3:05 AM Pacific time on 15 November, according to CyberPress. Engineers were automatically engaged five minutes later, at 3:10 AM. At 5:15 AM, AWS confirmed there were “no workarounds for impacted customers” for the authentication failures. The picture only started to improve at 5:20 AM, when engineers fully mitigated the authentication issue and most affected services recovered. An update two minutes later reported early recovery for IAM Identity Center and AWS Sign Up.
| Phase | Window (Pacific, 15 Nov) | Duration | What happened |
|---|---|---|---|
| 1. Identity failure | 3:05 AM to 5:20 AM | 2h 15m | IAM Identity Center sign-in failures; no workaround; sign-ups blocked |
| 2. Residual network failure | 5:20 AM to 6:21 AM | 1h 01m | Sign-ins fixed, but a subset of Network Load Balancers in one cell of use1-az1 kept dropping connections |
| Principal window (AWS summary) | 3:05 AM to 6:10 AM | 3h 05m | AWS’s own headline window for the disruption |
| Final all-clear | 6:53 AM | None | “Service is operating normally: [RESOLVED] Increased Connectivity issues in US-EAST-1” |
The table is our own reconstruction from the dashboard timestamps, and it shows something no single update spells out: the incident had two phases, an authentication failure followed by a residual network-connectivity problem. The authentication failure lasted 2 hours 15 minutes with zero workaround. The network tail ran for 1 hour 1 minute after the sign-in fix, which is about a third of the 3 hour 5 minute principal window. And the final all-clear at 6:53 AM came 32 minutes after the last mitigation at 6:21 AM, so the final resolution notice came 43 minutes after the end of AWS’s stated principal disruption window. Anyone counting the outage as “about three hours” is reading the headline, not the whole event.
AWS’s updates put the load-balancer fix at 6:21 AM Pacific. A 6:23 AM update confirmed recovery for most services and advised affected customers to create a new Network Load Balancer with targets spread across multiple Availability Zones. The final resolution notice followed at 6:53 AM, telling customers to retry any requests that failed during the outage.
The 11 services that went down
AWS listed 11 affected services. A major part of the disruption involved identity and access, which is why the blast radius looked the way it did:
| Service | What broke |
|---|---|
| AWS IAM Identity Center | Workforce sign-in to the AWS console and integrated applications failed |
| AWS Sign Up | New account sign-ups blocked |
| Kiro | AWS’s AI coding tool, cut off because it signs in through Identity Center; users on social sign-in were unaffected |
| Amazon Connect | Contact-centre platform, cut off through the same sign-in path |
| Amazon CloudWatch | Monitoring service listed among the affected services |
| AWS NAT Gateway | Network address translation listed among the affected services |
| Amazon Elastic Load Balancing | Load balancing listed among the affected services; a subset of Network Load Balancers kept dropping connections after sign-ins recovered |
| Amazon VPC Lattice | Service networking listed among the affected services |
| Amazon OpenSearch Service | Search and analytics service listed among the affected services |
| AWS Step Functions | Workflow orchestration listed among the affected services |
| AWS IoT Device Management | Device management listed among the affected services |
The Identity Center failure is the one that mattered most. Customers relying on it to reach the AWS Management Console could not authenticate at all, and that failure also cut off services that log in through Identity Center, including Amazon Connect and Kiro. Kiro users who sign in through social authentication methods were unaffected, which is a small but telling detail: the break was in the workforce identity path, not in the services themselves. The outage came days after Microsoft’s 12 to 13 November Microsoft 365 outage, which TechReview covered separately, but nothing in AWS’s updates links the two events.
The gaming services stayed up
For all the disruption, consumer impact was limited. Epic Games’ status page logged no incident for Fortnite or any of its online services on 15 November, AWS listed no games among the 11 affected services, and searches for player reports of Fortnite, Valorant, Roblox, PUBG or Twitch going down during the outage window turned up none, according to TalkEsport.
That is the full extent of what the reports establish. They show no outage for the named gaming services during this incident, and they do not explain why. The November outage broke sign-ins to the AWS console for customers relying on Identity Center. The available reports did not show an outage for the named consumer gaming services.
The contrast with last October is the point. The same region’s failure then knocked Fortnite and Roblox offline for most of a day. A year earlier, on 20 October 2025, a DNS problem with DynamoDB in US-EAST-1 lasted around 15 hours, according to AWS’s own post-event summary. Same region, opposite blast radius: in October 2026 the games went down, while in November the affected customers relying on Identity Center were unable to sign in to the console and integrated applications for 2 hours 15 minutes, and no outage was reported for the named gaming services. US-EAST-1 can fail in two completely different ways, and this time the customers who could not work were the ones administering AWS.
What AWS has not said
Three things are missing from AWS’s public updates, and all three matter. First, the company has not published a root cause. Its updates describe authentication and connectivity failures and stop there. They do not identify a failed component or say whether the authentication failure and the load-balancer failure shared one initiating event. Second, AWS has not attributed the outage to malicious activity or credential compromise. The incident was presented as an operational and availability problem, not a breach, and nothing in any source claims data was lost or stolen. CyberPress explicitly distinguishes this incident from the separate IAM credential-theft intrusion campaigns. Third, no affected-user count has been disclosed, so do not expect one in this article. No source gives a number, and Downdetector figures do not exist for this incident.
As of 16 November, no post-incident summary had appeared. Until one does, any explanation of what actually broke is speculation, and this article will not offer one.
What should I do if my AWS console was down?
If your console sign-in failed during the outage, AWS said there was no workaround for customers affected by the authentication failure, and the only action was to wait for the 5:20 AM fix, then retry whatever failed and check the Health Dashboard for updates.
- Retry any request that failed during the outage window, as AWS advised in its final notice.
- Check the AWS Health Dashboard before rebuilding anything, so a regional problem is not mistaken for a broken deployment.
- Do not rotate credentials or rebuild infrastructure mid-outage unless you have evidence of a separate problem.
- If load-balancer connections kept dropping, create a new Network Load Balancer with targets spread across multiple Availability Zones.
- Watch for AWS’s public post-incident summary before drawing any conclusions about what caused the failure.
These steps combine AWS’s own advice with general precautions. The retry instruction and the load-balancer advice come from AWS’s updates. The rest is what an affected customer can do when the console itself is unreachable: verify the scope first, change nothing under pressure, and wait for the vendor’s own account of the cause.
What to watch next
Four things are worth watching. First, whether AWS publishes a post-incident summary and what it says about the root cause. Second, whether AWS discloses a shared cause linking the authentication failure and the load-balancer cell failure, or confirms they were independent events that happened to overlap. Third, whether the Health Dashboard updates get revised, including the service list or the disruption window. Fourth, whether AWS says anything about Identity Center itself: customers relying on it use it to reach the console, and this outage showed what happens when that path fails. Until then, the facts of the AWS US-EAST-1 outage are 11 services, a 3:05 AM start, a two-phase recovery, and a region back to normal by 6:53 AM on a Sunday, with no cause named and no users counted.


