< ciso
brief />
Tag Banner

All news with #kubernetes tag

65 articles

Rackspace Spot Breach Exposes Customer Kubernetes Secrets

๐Ÿ” Rackspace has warned customers that attackers used leaked credentials to breach its Rackspace Spot Kubernetes platform on November 17, 2026. The intruders were able to impersonate legitimate users, create pods, and read Kubernetes Secrets in some customer clusters. Rackspace has revoked tokens, rotated keys, and urged affected customers to reissue API tokens and rotate stored credentials. The incident also caused an extended console outage before service was restored on November 21.
read more โ†’

CloudWatch EKS Adds Built-In Monitoring for Open Source Tools

๐Ÿ“Š Amazon CloudWatch now provides built-in monitoring for open source tools running on Amazon EKS clusters, including Karpenter, KEDA, vLLM, KServe, SGLang, Ray and Knative Serving. Monitoring starts automatically once OTel Container Insights is enabled, with no need to run Prometheus or Grafana yourself. Teams get dashboards for AI inference and autoscaling, with every metric tagged by pod, workload and node.
read more โ†’

AWS Secrets Store CSI Driver Adds AssumeRole Support

๐Ÿ” The AWS Secrets Store CSI Driver Provider now supports AssumeRole role-chaining, letting Kubernetes pods read secrets and parameters across AWS accounts. Teams simply name a target role in the SecretProviderClass, and the provider assumes it to fetch the data. This removes the need for per-secret resource policies and customer-managed KMS keys. The feature is available in all Regions where Amazon EKS and the provider are offered.
read more โ†’

GKE Adds In-Place Migration to Dataplane V2 in Preview

๐Ÿš€ Google Cloud has launched a preview of in-place migration for GKE Dataplane V2, letting teams move eligible VPC-native clusters from Dataplane V1 without building parallel clusters. The upgrade relies on eBPF and Cilium to deliver better scale, security, and observability. Rollout follows existing maintenance windows and upgrade strategies, with rollback available before finalization.
read more โ†’

Amazon EKS increases etcd size for larger clusters

๐Ÿ”ง Amazon EKS now provides a larger 24 GB etcd database (up from 16 GB) for 4XL and 8XL Provisioned Control Plane clusters on EKS version 1.34 or later. The etcd store holds cluster state, and the 50% capacity increase allows more Kubernetes objects, supporting larger pods, jobs, events, and custom resources. Existing eligible 4XL and 8XL clusters receive the upgrade automatically, and the 24 GB database is available in all commercial AWS Regions where Provisioned Control Plane is offered.
read more โ†’

Open AI-native Stack on Kubernetes for Production

๐Ÿงญ This post describes an open-source, Kubernetes-native AI stack that moves models from research into multi-tenant production. It highlights components contributed to the CNCF Sandbox โ€” including OpenRL, llm-d 1.0, KubeRay, Agent Substrate, and gVisor โ€” and shows how they address RL post-training, distributed inference, cache-aware routing, and secure agent execution. The article also summarizes benchmark results, operational benefits, and KubeCon NA 2026 demos.
read more โ†’

Mitigating protocol security gaps in multi-tenant Kubernetes

๐Ÿ”’ This post outlines an architecture for securing unencrypted or unauthenticated protocols (for example, ADB and WebRTC) in multi-tenant Amazon EKS clusters using three reusable components: namespace-scoped mTLS gateways, managed TURN via Amazon Kinesis Video Streams, and a JWT identity chain tied to an enterprise IdP. It assumes cert-manager, AWS Private CA, and Envoy Gateway are available and focuses on design choices and operational learnings rather than step-by-step deployment. The approach reduces attack surface, simplifies TURN operations by removing a custom token-exchange service, and preserves namespace isolation while ensuring auditability of user actions.
read more โ†’

Google donates agent runtime stack to CNCF

๐Ÿงญ Google is donating three open source projectsโ€”Agent Substrate, gVisor, and OpenRLโ€”to the Cloud Native Computing Foundation, and Agent Substrate has reached 1.0 and is generally available on GKE. These projects provide an open stack for running large-scale, stateful autonomous agents across CPUs, GPUs, and TPUs by decoupling agent state from compute, sandboxing untrusted code, and orchestrating reinforcement learning post-training loops. The move aims to enable neutral governance, broader community contributions, and production-ready isolation and efficiency for agentic workloads.
read more โ†’

Amazon EKS Auto Mode adds Gateway API support

๐Ÿ› ๏ธ Amazon EKS Auto Mode now supports the Kubernetes Gateway API, enabling a standard, portable way to define traffic routing to applications. Customers can use Gateway and HTTPRoute resources while EKS Auto Mode provisions and manages underlying AWS infrastructure automatically. Features include HTTP and gRPC routing, cross-namespace traffic, weighted splitting, header-based rules, and automatic TLS via AWS Certificate Manager. Existing Ingress resources continue to operate alongside Gateway API resources.
read more โ†’

llm-d 1.0: Open Kubernetes Stack for LLM Inference

๐Ÿงญ Today at the PyTorch Conference, Google and partners released llm-d 1.0, an open-source, Kubernetes-native stack for distributed LLM inference developed in the CNCF Sandbox. The project unifies PyTorch serving engines (vLLM, SGLang) and Kubernetes primitives to route, cache, disaggregate, and scale model inference across mixed TPU and GPU fleets. It provides prefix-aware routing, KV cache offload, prefill/decode disaggregation, async processing, and time-slicing to improve latency and throughput for multi-step agent and RL workloads.
read more โ†’

Amazon EKS Distro Adds Kubernetes 1.37 Support

๐Ÿš€ Amazon EKS and EKS Distro now support Kubernetes version 1.37. This release lets you create new clusters or upgrade existing ones via the EKS console, eksctl, or infrastructure-as-code tools. Kubernetes 1.37 promotes the Metrics API to GA, graduates Dynamic Resource Allocation device taints and tolerations to GA, and enables HPA scale-to-zero by default. EKS 1.37 is available in all Regions where EKS operates, with EKS Distro images published to ECR Public Gallery and GitHub.
read more โ†’

GKE CPU Startup Boost for Faster Pod Starts

๐Ÿš€ GKE introduces CPU startup boost in preview, integrated into the Vertical Pod Autoscaler (VPA), to temporarily elevate CPU allocation during container initialization and then in-place scale it back to steady-state levels without restarting containers. The feature leverages Kubernetes In-place Pod Resize (IPPR) and operates across admission, startup, and unboosting phases. It targets CPU-intensive startup workloadsโ€”such as Java JVMs, Node.js, and Python/AI microservicesโ€”so teams can avoid overprovisioning while reducing cold starts and costs. Available on GKE 1.36.0-gke.4447000+ for Standard and Autopilot clusters.
read more โ†’

EKS Auto Mode adds advanced node tuning options

๐Ÿš€ Amazon EKS Auto Mode now supports advanced node tuning via the NodeClass resource, enabling kubelet, Linux kernel sysctl, and hugepage configuration directly through the Kubernetes API. advancedCompute lets you set user namespaces for rootless builds, adjust eviction thresholds and log rotation, raise network and ARP cache limits, and pre-allocate 2Mi or 1Gi hugepages for HPC and low-latency workloads. EKS Auto Mode validates and applies settings at node boot and preserves them across scaling and upgrades, with no custom AMIs, EC2 launch templates, or privileged DaemonSets to manage.
read more โ†’

Compromised Identity Leads to Broad DevOps and Cloud Access

๐Ÿ”’ Microsoft DART investigated an incident where the Storm-3068 actor turned a self-service password reset into persistent access across Azure DevOps, development pipelines, and Kubernetes. The actor used legitimate identity and cloud services to enumerate repositories, create malicious pipelines to harvest kubeconfig files, and deploy remote access tools like Atera and Chisel. DART worked with the customer to contain the intrusion, reconstruct activity from audit logs and Git history, and provide remediation guidance to reduce future identity-driven risks.
read more โ†’

GKE Adds Native Prometheus Metrics for Autoscaling

๐Ÿ”” Google Cloud announced built-in Prometheus metrics processing for GKE, enabling HPA to use PromQL queries directly via Google Managed Service for Prometheus. This removes the need for third-party adapters, reduces latency, and simplifies autoscaling configuration. The controller runs in the control plane and only deploys pods on nodes when PromQL metrics are actively requested, minimizing resource use. The feature is in preview with plans to add self-hosted Prometheus support before GA.
read more โ†’

GKE Adds Native Scale-to-Zero Capabilities

๐Ÿš€ GKE 1.37 introduces native scale-to-zero so workloads can fully scale down to zero replicas and stop consuming compute while idle. The feature uses HPA with the new AutoscalingMetric CRD and KEP-2021 support for minReplicas: 0 to wake workloads based on external signals such as Pub/Sub or Cloud Monitoring. Capacity buffers provide pooled warm capacity to eliminate cold-start latency and balance cost with instant responsiveness.
read more โ†’

Amazon EMR on EKS adds IPv6 support

๐Ÿš€ Amazon announces that Amazon EMR on EKS now supports running Spark and Flink workloads on IPv6 Amazon EKS clusters. This enables teams to use the larger IPv6 address space to scale high-executor analytics jobs without IPv4 workarounds. Support starts with EMR releases emr-7.14.0 and emr-spark-8.0.0 and is available in regions where IPv6 EKS clusters are offered.
read more โ†’

AWS PCS adds GRES, hardware topology, and MIG support

๐Ÿ”ง AWS Parallel Computing Service (AWS PCS) now supports custom Generic Resource (GRES) settings, hardware topology configuration, and NVIDIA Multi-Instance GPU (MIG), and upgrades to Slurm 26.05. These additions let Slurm autodetect NUMA domains as sockets, configure GPU core affinity and GPU interconnects, and enable finer control over task placement for HPC and ML workloads. Custom GRES and topology settings let administrators declare accelerators, partition GPUs with MIG, or override defaults for specialized scheduling.
read more โ†’

AWS Private CA connectors arrive in GovCloud

๐Ÿ”’ AWS Private Certificate Authority (AWS Private CA) now offers the AWS Private CA Connector for Kubernetes as a managed Amazon EKS add-on and the AWS Private CA Connector for Active Directory in AWS GovCloud (US-East) and (US-West). These additions simplify certificate automation for government workloads, integrating with cert-manager to provision and renew TLS certificates in Kubernetes and enabling AD-based automatic issuance for domain-joined objects. The service secures private keys using FIPS 140-3 Level 3 HSMs.
read more โ†’

Amazon EKS Argo CD now supports custom configuration

๐Ÿ”ง The Amazon EKS Capability for Argo CD now accepts a standard argocd-cm ConfigMap in your cluster to enable custom configuration. This managed GitOps continuous delivery experience can be tuned to your teamโ€™s workflows, including custom health checks for Custom Resources, UI banner content, and resource watch/compare behavior. AWS applies settings configured the same way as upstream Argo CD, and built-in health checks for AWS Controllers for Kubernetes (ACK) and kro resources are included.
read more โ†’