status.nebius.com
AI cloud & inference platform
All Systems Operational
The root cause was found and this incident has been resolved.
We are currently experiencing a partial unavailability of our Compute services, which may affect some of your workloads. Our team is actively investigating the issue and working towards a resolution. We will provide further updates as soon as more information becomes available.
This incident has been resolved.
Root cause is located and fixed. We keep monitoring the resolution.
We are investigating an issue with IAM Tenants Status change. Customers may not be able to resume their accounts from the suspended state.
This incident has been resolved.
We are currently investigating this issue.
This incident has been resolved.
The team has identified the root cause of the issue and is currently applying a fix. We expect the incident to be fully mitigated within two hours.
Users might be experiencing a high rate of 5xx errors on upload requests and elevated latency on read requests.
This incident has been resolved.
The error rate has dropped to zero. Requests to create, update, start, and stop SkyPilot clusters in eu-north1 are completing successfully again. We are continuing to investigate the root cause and are monitoring the service closely. Instances that were already running have not been affected at any point. We will provide a further update as the investigation progresses.
We are investigating an issue affecting Managed Service for SkyPilot in eu-north1. Requests to create or update SkyPilot clusters, as well as start and stop requests, are failing with HTTP 503 errors. Instances that are already running are not affected and continue to operate normally.
This incident has been resolved.
The fix has been deployed and the team is looking into the instrumentation for any signs of any remaining abnormal behaviour.
In some cases VMs may fail to create or metadata inside may not be available. The team has identified the issue and is working on deploying the fix.
This incident has been resolved.
We are continuing to monitor for any further issues.
The issue may have caused consequential impact to VPC functionality of VMs that are not Managed Service for Kubernetes worker nodes.
Worker nodes of Managed Service for Kubernetes clusters created after 15:13 CET Sep 1 2026 could experience problems connecting to/from other nodes in the same cluster. The root cause is understood and mitigation is underway.
[https://nebius.com/blog/posts/incident-post-mortem-analysis-us-central1-service-disruption-on-august-19](https://nebius.com/blog/posts/incident-post-mortem-analysis-us-central1-service-disruption-on-august-19)
This incident has been resolved.
Most critical issues have been resolved and the region is operational, though a subset of nodes remains unavailable.
The Managed Service for Kubernetes is nearly fully restored, though a subset of nodes remains unavailable.
Most nodes are operating normally. The Managed Service for Kubernetes® is still degraded, but the situation is improving.
Most nodes are operating normally. The Managed Service for Kubernetes® is experiencing a partial outage, and the team is actively working to restore it.
Services are still gradually continuing to recover. Many of the virtual machines are already running, but not all of them.
Services are still gradually continuing to recover. Some virtual machines are already running, but not all of them.
Services are gradually continuing to recover.
The root cause was fixed, and the situation is stabilizing. The services are recovering.
We are still experiencing temperature issues in the us-central1. Team is working on it
The root cause has been identified as overheating hardware. GPU performance is impacted, and some nodes may be unavailable.
We are currently investigating this issue.
Some newly registered users may have experienced issues logging in to the management console between 13:00 UTC on 17 August 2026 and 14:00 UTC on 18 August 2026.
This incident has been resolved.
We are currently investigating this issue.
This incident has been resolved.
According to our metrics, the situation with VPC has stabilised.
We are investigating an issue with external VPC connectivity
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
We are continuing to investigate this issue.
Due to issue with internal services the creation of new VM and other operations may be affected. We are investigating the issue.
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
The issue has been identified and a fix is being implemented.
We are currently investigating this issue.
The eu-north1 region experienced network issues that may have affected the operation of several services. The issue has been resolved, and all services are operating normally.
This incident has been resolved.
We are currently investigating this issue.
Managed SkyPilot OIDC Login was unavailable since 2 of July. The problem has been identified and resolved.
This incident has been resolved.
A fix has been implemented and we are monitoring results.
We are continuing to investigate this issue.
We are continuing to investigate this issue.
We are currently experiencing partial unavailability of our Monitoring, Object Storage, Token Factory services. Our team is actively investigating the issue and working towards a resolution. We will provide further updates as soon as more information becomes available.
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
The issue has been identified and a fix is being implemented.
The Object Storage service in the US-CENTRAL1 region has been fully restored and is operating normally. The root cause has been identified, and corrective measures are in place.
The Object Storage service in the US-CENTRAL1 region has recovered and is operating normally. We are monitoring to confirm stability. Root cause analysis is ongoing
We are continuing to investigate this issue.
We are investigating an outage affecting the Object Storage service in the US-CENTRAL1 region.
This incident has been resolved.
One InfiniBand switch in the fabric-3 cluster is down.
Connectivity issues are gone now. Intermittent failure is being investigated in the background with the uplink provider, while the Nebius Netowrking team continues to monitor the link closely.
We have noticed that there have been problems with the connection latency to external addresses in eu-north1.
This incident has been resolved.
We're currently investigating an issue affecting Compute instances startup. Some requests to create or start instances may fail. Already-running instances are not affected. Our engineering team is actively investigating.
Some of public IPs became unreachable.
This incident has been resolved.
The issue has been resolved, and affected services have recovered. The situation continues to be monitored to ensure service stability.
An issue with Internet connectivity in the data center has been detected. Impact: - Logs and metrics from virtual machines are not being delivered. - Object Storage is currently unavailable from internet. - Container registry is currently unavailable from internet. - Additional services relying on external connectivity may be affected. The issue is under investigation.
This incident has been resolved.
Public IPv4 address capacity has been restored in eu-north1, and new public IPv4 addresses can be allocated to workloads again. We are monitoring the service to confirm full recovery before resolving this incident.
We have identified the cause: the pool of available public IPv4 addresses in eu-north1 is currently exhausted, which is why new public IPv4 addresses cannot be allocated to workloads. We are reclaiming and reallocating addresses to restore allocation capacity. Workloads that already have a public IPv4 address assigned remain unaffected. We will provide a further update once new allocations are succeeding.
We are investigating an issue in eu-north1 affecting Virtual Private Cloud (Networking). New public IPv4 addresses cannot currently be allocated to workloads. Customers may be unable to assign a public IP when creating or modifying resources — for example, attaching a public IP to a virtual machine or load balancer in this region. Resources that already have a public IPv4 address assigned are not affected and continue to operate normally. We will provide an update as we learn more.
Several servers were rebooted.
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
The issue has been identified and a fix is being implemented.
This incident has been resolved.
We are investigating reduced InfiniBand fabric performance affecting GPU workloads in the uk-south1 region. Customers may observe lower throughput and slower distributed training jobs.
Issue resolved 21:10.
We are continuing to investigate this issue.
Nebius console is unavailable
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
The issue has been identified and the fix is being implemented
Managed Kubernetes operations can take longer than expected, leading to delays or timeouts. Deletion operations are particularly impacted. The team is currently investigating the issue.
The external provider has restored operations. All models are now available.
The following serverless models are unavailable due to external provider downtime. Qwen/Qwen3-235B-A22B-Thinking-2507-fast Qwen/Qwen3-Next-80B-A3B-Thinking-fast Qwen/Qwen3.5-397B-A17B-fast deepseek-ai/DeepSeek-V3.2-fast openai/gpt-oss-120b-fast
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
We are currently investigating this issue.
This incident has been resolved.
Most of Data Lab functionality is restored. Data filtering functionality is disabled until future updates.
We are continuing to investigate this issue.
We are currently investigating this issue.
Since 8:30 back to normal.
The issue has been identified and a fix is being implemented.
This incident has been resolved.
Impact is mitigated and we are monitoring the situation
We are continuing to investigate this issue.
Users may experience issues when working with the following serverless models in the us-central1 region: - Qwen/Qwen3.5-397B-A17B - MiniMaxAI/MiniMax-M2.5 We are currently investigating the issue and will provide updates as more information becomes available.
During the night we experienced issues with storage service. While writing for data was not affected (just delayed), but reading logs, accessing metrics and audit trails in console was not possible while the incident was in active stage.
This incident has been resolved.
The problem was fixed, and we are monitoring the results.
We are noticing issues with some customers viewing metrics for Dedicated Endpoints in Token Factory. We are currently investigating this issue.
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
The issue has been identified and a fix is being implemented.
We are currently investigating this issue.
A fix has been implemented and we are monitoring the results
We are continuing to investigate this issue.
We are currently experiencing a partial unavailability of our mk8s services, which may affect some of your workloads. Our team is actively investigating the issue and working towards a resolution. We will provide further updates as soon as more information becomes available.
We had a problem with metrics ingestion which resulted in storage metrics not being ingested during the following intervals: - eu-west: from 14:50 CET to 19:05 CET - me-west1: from 17:00 CET to 19:05 CET - eu-west1: from 15:50 CET till 19:05 CET Our team has resolved this issue, new metrics are being ingested, but, unfortunately, data for that period is lost
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
We are currently investigating this issue.
This incident has been resolved and all services and clusters should be operational now.
Maintenance has been completed, and we do not expect any further power-related issues. We are now recovering services and closely monitoring system status.
We are continuing to investigate this issue.
We are continuing to investigate this issue.
We are currently experiencing another service disruption due to ongoing on-site maintenance. Our engineers are actively working to stabilize the affected clusters and maintain service throughout the maintenance period.
We have identified failed switches and nodes. All major services were recovered, we are now collecting faulty instances and recovering them.
We are continuing to work on a fix for this issue.
We are continuing to work on a fix for this issue.
We are continuing to work on a fix for this issue.
During planned maintenance we experienced unexpected issues with power supply. We have identified the faulty circuits and already bringing nodes that are not part of planned maintenance back to work.
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
The issue has been identified and a fix is being implemented.
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
The issue has been identified and a fix is being implemented.
Several racks failed.
This incident has been resolved.
The root cause was identified, failed workloads restored.
We are continuing to investigate this issue.
We’re experiencing power issues in the eu-north1 region. Some resources (virtual machines, network disks and filestores) may be unavailable.
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
The issue has been identified and the team is implementing a fix.
We are currently investigating the issue
This incident has been resolved.
Batch inference is temporarily unavailable.
This incident has been resolved.
A fix has been implemented and we are monitoring the results.
Some Managed Kubernetes k8s LoadBalancers transitioned to the Pending state and could be temporarily unavailable. We are actively working on mitigation and clarifying the scope of the impact.
This incident has been resolved.
A faulty IB switch was rebooted and sucesfully running. The team monitors the situation
The issue has been identified and a fix is being implemented.
The connectivity is fully restored.
Links are back up, we are monitoring the equipment and bandwidth saturation.
The switch in question has been cycled.
One of IB switches has failed. DC ops are working on restoring it.