tag: Kubernetes · 169 items
- Platform/SRE — Learn: Volcano v1.16 introduces namespace-scoped hierarchical queues and fairshare scheduling, worth evaluating if you run batch or ML workloads on Kubernetes; no EOL pressure or breaking changes signaled.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Kubernetes is formalizing AI workload conformance standards under SIG Architecture, which shapes how AI inference and training workloads will be certified to run on Kubernetes; no operational change needed today, but worth tracking as conformance requirements may influence cluster configuration choices for AI workloads.
- CI/CD — Skip
- Leader — Learn: Kubernetes’s elevation of AI Conformance to a formal SIG subproject signals the ecosystem is converging on standards for AI workloads, useful context when evaluating long-term platform strategy for AI infrastructure.
- Platform/SRE — Learn: An early-stage self-hosted PaaS layer on top of Kubernetes worth watching as an IDP building block, but with only 115 stars and no enrichment signals it’s too immature to evaluate for production adoption today.
- CI/CD — Skip
- Leader — Learn: An emerging open-source self-hosted PaaS option relevant to internal developer platform strategy; no concrete adoption decision warranted yet given its early community size.
- Platform/SRE — Learn: The WG targets exactly the repeated per-cluster workload assessments operators do today — node-drain tolerance, removed-API dependencies, over-privileged containers. Worth tracking as the spec develops, but the deliverable doesn’t exist yet so there’s nothing to act on.
- CI/CD — Skip
- Leader — Learn: A Kubernetes-blessed workload conformance standard could eventually reduce per-environment onboarding overhead and inform the org’s golden-path requirements, but the working group is newly chartered with no published spec yet — no planning decision warranted now.
- Platform/SRE — Plan: etcd backs every Kubernetes control plane, and a breaking change in a patch release demands a CHANGELOG review before upgrading; review the upgrade guide and validate cluster behavior in staging before rolling this out to production — no forced deadline yet.
- CI/CD — Skip
- Leader — Skip
- Signals: breaking-change flagged
- Platform/SRE — Plan: etcd v3.6.16 carries a flagged breaking change in a patch release, which is atypical — review the changelog and upgrade guide before applying, since etcd is the Kubernetes control-plane backing store and an unexpected breakage can bring down cluster state. No hard deadline, but treat this as a deliberate upgrade rather than a routine patch rollout.
- CI/CD — Skip
- Leader — Skip
- Signals: breaking-change flagged
- Platform/SRE — Plan: Unusual for a patch release, v3.5.35 carries a breaking-change flag; review the etcd CHANGELOG and upgrade guide before applying this to any cluster running etcd 3.5.x to avoid unexpected behavior in the control-plane data store.
- CI/CD — Skip
- Leader — Skip
- Signals: breaking-change flagged
- Platform/SRE — Learn: A community toolkit covering VMware-to-K8s migration, Compose translation, and Alibaba Cloud HA setups — worth bookmarking for teams planning a VMware exit, but it carries no deadline, no enrichment signals, and only 52 stars, so there’s nothing to act on today.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Describes an architecture pattern for GPU multi-tenancy using Kamaji (hosted control planes) and HAMi, worth evaluating if your org runs shared GPU clusters — no GA migration pressure, but a useful design reference.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: New console visibility into MIG instance distribution across zones and machine types — useful context for capacity planning and troubleshooting, but no operational change required.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: CVE-2026-95839 is a security fix in the container runtime underlying every Kubernetes node, but it is not KEV-listed and carries no public exploitation signal, so this does not meet the Act bar. Plan to upgrade containerd to 2.4.2 across node pools this quarter.
- CI/CD — Skip
- Leader — Skip
- Signals: containerd 2.4 EOL 2027-05-16 · CVE-2026-95839 — CISA KEV: not listed, EPSS n/a, reported by 2 collected sources
- Platform/SRE — Learn: Useful framing on how GPU-bound and bursty AI workloads strain the classic HPA/VPA/cluster-autoscaler division; no version change or deadline, but informs how you’d architect autoscaling for AI teams onboarding to the cluster.
- CI/CD — Skip
- Leader — Learn: Useful context for platform strategy as teams push AI/ML workloads onto shared Kubernetes; no near-term decision or cost/vendor change implied, but shapes how you’d evaluate autoscaling capabilities when reviewing cluster investments.
- Platform/SRE — Learn: Practical case study on running a stateful, multi-component workload (workers, webhook processors, PostgreSQL, Redis) on EKS; useful pattern reference for teams operating similar architectures, but no action required.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Skip
- CI/CD — Skip
- Leader — Learn: A practitioner account of moving off Kubernetes with a positive outcome — useful background reading when evaluating whether Kubernetes complexity is justified for a given org’s scale, but no strategic decision is required.
- Platform/SRE — Learn: Useful mental model for diagnosing Kueue admission delays — separating queue admission from Pod scheduling is a non-obvious distinction worth internalizing. The Headlamp plugin is 0.1.0-alpha, so no adoption yet, but the investigation path applies with kubectl today.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Speculative design discussion about a hypothetical Kubernetes 2.0 — no actual release, no deadline, no actionable change to running clusters, but shapes thinking about future cluster architecture.
- CI/CD — Skip
- Leader — Learn: High-level architectural speculation relevant to long-term platform strategy, but no concrete product decision or vendor action required today.
- Signals: major release (2.0)
- Platform/SRE — Learn: LLM-D is an emerging Kubernetes-native distributed inference project worth tracking as AI workloads increasingly land on shared clusters; no GA release or concrete migration action yet.
- CI/CD — Skip
- Leader — Learn: Signals a maturing pattern for running LLM inference on Kubernetes infrastructure; worth monitoring as a potential build-vs-buy input for AI serving strategy, but no decision point yet.
- Platform/SRE — Learn: An open-source PaaS layer on top of Kubernetes that adds DNS/SSL/Helm chart management; worth evaluating for teams building an internal developer platform, but no production signal or GA maturity indicated.
- CI/CD — Skip
- Leader — Learn: Illustrates the cost gap between managed PaaS (Heroku, Render, Fly) and raw VPS hosting; relevant context if evaluating internal platform strategy or PaaS spend, but no actionable decision required now.
- Platform/SRE — Learn: Flox brings Nix-based reproducible environments to Kubernetes pods, an interesting pattern for standardizing dev/test workloads on shared clusters — worth evaluating but no GA urgency or operational deadline.
- CI/CD — Learn: Nix-based environments in Kubernetes could improve build reproducibility for pipeline workloads, but this is an exploratory pattern with no concrete migration or deadline attached.
- Leader — Skip
- Platform/SRE — Learn: Practical technique for air-gapping or mirroring the pause container image — relevant to clusters that need to avoid external registry dependencies, but no deadline or active issue drives immediate action.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: ingress-nginx is one of the most widely deployed Kubernetes ingress controllers; its retirement means planning a migration to an alternative (Gateway API, Traefik, Emissary, etc.) before it goes unmaintained. No hard deadline surfaced in signals, so schedule migration evaluation this quarter.
- CI/CD — Skip
- Leader — Plan: Organizations standardized on ingress-nginx will need to choose and adopt a replacement ingress strategy; evaluate Gateway API or alternative controllers and factor the migration into the next infrastructure planning cycle.
- Signals: deprecation mentioned (no explicit date found)
- Platform/SRE — Learn: A practical walkthrough of using Squid as a Kubernetes egress proxy — useful design reference if evaluating egress filtering patterns, but no deadline or breaking change to act on.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: A practitioner’s conceptual framing of Kubernetes may sharpen mental models for platform engineers, but nothing here requires operational change.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Extreme-scale GKE engineering case study surfaces real control-plane and networking constraints worth tracking as clusters grow; no operational change needed now.
- CI/CD — Skip
- Leader — Learn: Illustrates how Google scales managed Kubernetes for massive AI workloads — useful context when evaluating GKE capacity or large-cluster architecture decisions.
- Platform/SRE — Learn: A new opinionated provisioning layer for k3s on Hetzner that bundles add-ons and GitOps-style deployments; worth evaluating if cost-driven teams are exploring Hetzner as a cheaper alternative to AWS/GCP managed K8s, but no existing infrastructure to change today.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Databricks-scale post-mortem on Kubernetes load balancing patterns; useful for informing how you’d design or tune traffic distribution on large clusters, but no operational change required.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: An early-stage tool worth evaluating if you run bare-metal or on-prem Kubernetes — it exposes BMC power control and hardware health as native K8s resources with Prometheus metrics. No action needed today; 54 stars suggests pre-production maturity, but the pattern is worth watching for hybrid infrastructure designs.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: If you run Azure Arc-enabled Kubernetes clusters, plan migration from the Azure Key Vault Secrets Provider Extension to the Azure Key Vault Secret Store Extension before the hard retirement date of October 9, 2027. One year is enough runway to schedule the migration, but start scoping which clusters are affected now.
- CI/CD — Skip
- Leader — Plan: Azure is retiring a secrets-integration extension for Arc-enabled Kubernetes with a clear replacement in the same ecosystem; ensure teams using Arc clusters have this migration on their roadmap before October 9, 2027, as there is no stay-on-current option beyond that date.
- Signals: deprecation/EOL deadline mentioned: October 9, 2027
- Platform/SRE — Learn: Reflective essay on how internal platforms often reinvent Kubernetes primitives — useful framing for evaluating whether your IDP is accruing accidental complexity.
- CI/CD — Skip
- Leader — Learn: Offers a mental model for the build-vs-buy question on internal developer platforms and the hidden costs of reimplementing orchestration primitives.
- Platform/SRE — Learn: Shapes how platform engineers think about structuring compliance evidence and control ownership on shared Kubernetes infrastructure, particularly for teams operating in EU-regulated environments — no operational change required today.
- CI/CD — Skip
- Leader — Learn: Useful framing for leaders clarifying accountability boundaries between platform and application teams under EU regulatory mandates, but contains no new deadlines or decisions — organizational context only.
- Platform/SRE — Learn: A new desktop Kubernetes management client worth bookmarking if you prefer a native GUI over kubectl or Lens, but no maturity signals, no GA confirmation, and no reason to change existing tooling today.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: K3k offers a Kubernetes-in-Kubernetes approach for isolated tenant clusters without full infrastructure cost — worth evaluating if you’re exploring multi-tenant platform patterns. No GA signal or deadline in the enrichment, so no action required today.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: GA feature worth adopting for teams using AWS Network Firewall with EKS/ECS — wildcard patterns like app=payments-* can replace per-variant rules, reducing rule sprawl in dynamic container environments. Evaluate and consolidate existing container attribute rules this quarter.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: A novel proof-of-concept running a Kubernetes control plane in the browser via WASM — interesting for understanding K8s internals and potential future dev/test tooling, but no production relevance today.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Anyscale on AKS is now GA — worth noting as a managed option for distributed Python/Ray workloads that deploys onto existing AKS clusters, but no operational change required unless your org is evaluating Ray.
- CI/CD — Skip
- Leader — Learn: Anyscale’s GA availability on Azure is a new build-vs-buy data point for orgs running distributed ML or data workloads on AKS; relevant if evaluating Ray adoption or comparing against self-managed Ray on GKE/EKS.
- Signals: GA announcement
- Platform/SRE — Plan: Kubernetes 1.35 (currently GA, EOL 2027-02-28) defaults failCgroupV1 to true, meaning kubelet refuses to start on cgroup v1 nodes without an explicit override. Audit all Linux nodes for cgroup version and complete migration to cgroup v2 before upgrading any cluster to 1.35 or later.
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.31 is past EOL (2025-11-11, 330d ago) · Kubernetes 1.25 is past EOL (2023-10-27, 1076d ago) · Kubernetes 1.35 EOL 2027-02-28 · deprecation mentioned (no explicit date found)
- Platform/SRE — Learn: Explains a gap in how Kubelet surfaces inode exhaustion — it alerts only when already in critical territory, not proactively. Worth factoring into NodeFilesystemFilesFillingUp alert tuning and inode-capacity headroom planning on worker nodes.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Public preview of AKS running directly on bare metal Ubuntu could simplify on-prem Kubernetes for GPU/AI workloads; worth evaluating, but pre-GA status means no production adoption yet.
- CI/CD — Skip
- Leader — Skip
- Signals: pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Learn: Pre-GA release of the Cilium CNI; notable upcoming changes include a new LB IPAM IP-sharing annotation for mixed-pod services and a fix for BGP route advertisement racing BPF attachment on node reboot — worth tracking before the GA upgrade cycle.
- CI/CD — Skip
- Leader — Skip
- Signals: deprecation mentioned (no explicit date found)
- Platform/SRE — Plan: Kubernetes v1.34 GA swap support with NVMe backing is worth evaluating for memory-dense node pools this quarter; assess enabling swap on nodes running agentic AI or sandboxed workloads to reduce OOM kills and improve pod density without fleet-wide urgency.
- CI/CD — Learn: The benchmark covers CI/CD kernel build workloads as a validated use case, but enabling swap on nodes is a platform configuration decision — useful context for understanding build pod resource headroom if your CI runs on shared Kubernetes clusters.
- Leader — Plan: Up to 3× pod density on existing hardware is a material FinOps opportunity worth adding to the next infrastructure planning cycle; evaluate whether adopting Kubernetes v1.34 swap + NVMe changes the cost model for memory-intensive AI and build workloads in your standard platform stack.
- Signals: GA announcement
- Platform/SRE — Plan: Kubernetes 1.37 is now GA on EKS with notable promotions (Metrics API to GA as metrics.k8s.io/v1, DRA device taints to GA). Plan an upgrade evaluation this quarter; no forced-upgrade deadline exists yet, but staying current reduces future skew risk.
- CI/CD — Skip
- Leader — Skip
- Signals: GA announcement
- Platform/SRE — Learn: Preview status caps this at Learn, but worth tracking: the Ubuntu2604 SKU will eventually replace older LTS node pools; plan evaluation for when it reaches GA before Ubuntu 22.04 support ages out.
- CI/CD — Skip
- Leader — Skip
- Signals: Ubuntu 26.04 EOL 2031-05-29 · pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Learn: Explores a design philosophy shift from blocking admission controllers to permissive guardrails — worth reading to inform how you configure OPA/Gatekeeper or Kyverno, but no operational change required today.
- CI/CD — Skip
- Leader — Learn: Shapes thinking on internal developer platform standards and how policy enforcement affects developer experience — useful context for platform strategy discussions.
- Platform/SRE — Plan: ARC runs on Kubernetes clusters you operate; the 0.15.0 improvements to runner scale-set reliability during cluster upgrades are worth evaluating and scheduling for adoption this quarter. No deadline or security fix in the signals.
- CI/CD — Plan: If you operate self-hosted GitHub Actions runners via ARC, 0.15.0 brings reliability and observability gains for large runner fleets — worth scheduling an upgrade to reduce disruption during runner or cluster maintenance windows.
- Leader — Skip
- Platform/SRE — Plan: GKE Standard’s GA increase to 512 pods per node (up from 256) doubles workload density but requires a /22 Pod CIDR block for nodes in the 257–512 range — evaluate whether your cluster IP address plan supports this before enabling denser node configurations.
- CI/CD — Skip
- Leader — Learn: GKE’s doubled pod-density ceiling may shift cluster sizing and cost economics for teams running dense workloads; the new X5 48TB memory tier is worth noting only if the org operates SAP HANA at scale and previously hit the 32TB single-node ceiling.
- Signals: GA announcement
- Platform/SRE — Plan: A new GA EKS managed add-on that replaces round-robin LB with real-time GPU-signal-aware routing (KV cache, queue depth, LoRA residency) for HyperPod clusters — evaluate adopting this quarter if your team runs SageMaker HyperPod inference workloads on EKS.
- CI/CD — Skip
- Leader — Learn: AWS’s inference gateway pattern (single endpoint, multi-model routing, hardware-aware scheduling) signals a maturing managed alternative to self-built vLLM fleet management — worth tracking as it informs the build-vs-buy framing for LLM serving infrastructure.
- Platform/SRE — Learn: New GA capability that could influence how data teams use EKS clusters — interactive Spark sessions run as pods with IAM-scoped roles, worth understanding for capacity planning and RBAC design.
- CI/CD — Skip
- Leader — Learn: Signals a maturing managed Spark-on-Kubernetes story from AWS; relevant when evaluating whether to consolidate data engineering workflows onto existing EKS infrastructure rather than separate EMR clusters.
- Platform/SRE — Learn: A community interview covering SIG Apps’ priorities around workload resilience, StatefulSet/Deployment evolution, and AI workload support — useful context on where Kubernetes primitives are headed but no operational change required today.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Public preview of AKS Flex Nodes enables hybrid and edge hosts as worker nodes connected to an Azure-managed control plane — worth evaluating if you run edge workloads, but pre-GA caps this at Learn.
- CI/CD — Skip
- Leader — Skip
- Signals: pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Learn: Beta (pre-GA) cap applies; the default-on Unused condition on PVCs is worth knowing for future storage-hygiene automation and cost-reduction tooling, but requires no change to running clusters today.
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.37 EOL 2027-10-28 · pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Learn: Beanstalk Cluster Mode is EKS under the hood with AWS managing the control plane — worth understanding as an abstraction layer above raw EKS, but platform engineers already running EKS have no action to take and no deadline.
- CI/CD — Learn: A new official Beanstalk GitHub Action ships alongside this feature, relevant only to teams already in the Beanstalk ecosystem; no deprecation or deadline pressure makes this worth monitoring rather than scheduling.
- Leader — Learn: Cluster Mode’s shared-infrastructure model could reduce per-app compute cost for orgs with many small services on Beanstalk, but there is no pricing change or strategic forcing function to act on — file for the next platform golden-path review.
- Platform/SRE — Plan: Kubernetes v1.37 GA ships emptyDir permission modes (including sticky bit) and bind mount options (noexec, nosuid, nodev) — schedule evaluation and adoption as part of your 1.37 upgrade cycle to harden shared writable volumes; no forced deadline until EOL 2027-10-28.
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.37 EOL 2027-10-28
- Platform/SRE — Learn: Practical integration guide for OpenBao (the Linux Foundation Vault fork) backed by a self-healing CloudNativePG instance — worth evaluating as a vendor-lock-in-free secrets management pattern, but no deadline or operational trigger to act on now.
- CI/CD — Skip
- Leader — Learn: Illustrates a viable open-source secrets management stack (OpenBao + CloudNativePG) that avoids HashiCorp vendor lock-in — useful context if the org is reassessing its Vault strategy, but this is a tutorial with no new strategic event attached.
- Platform/SRE — Learn: Beta graduation means the feature is now more stable and worth evaluating for latency-critical workloads using NUMA-aligned CPU/memory, but it remains disabled by default and pre-GA status caps this at Learn — no action needed until GA.
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.37 EOL 2027-10-28 · Kubernetes 1.36 EOL 2027-06-28 · pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Learn: Beta graduation means the MemoryQoS feature gate is on by default in v1.37 kubelets running cgroup v2, though safe defaults mean no throttling or reservation unless explicitly configured — worth understanding before upgrading clusters to v1.37.
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.37 EOL 2027-10-28 · pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Learn: CBT for CSI block volumes reached Beta in Kubernetes 1.33, with a breaking Alpha→Beta migration (v1alpha1 CRD removed; manifests and client code must be updated to cbt.storage.k8s.io/v1beta1). Still pre-GA so no production adoption urgency, but worth tracking if you ran the Alpha.
- CI/CD — Skip
- Leader — Skip
- Signals: pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Plan: Cilium 1.20 is a GA minor release of a CNI/service-mesh stack platform teams operate directly; new ENI IPAM for IPv6 and Gateway API ExternalAuth are worth evaluating for adoption this quarter, and the TCPRoute/UDPRoute additions may change how non-HTTP traffic routing is architected on the cluster.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: Notable operational improvements: concurrent status computation eliminates multi-minute delays and the resync period drops from 1 hour to 3 minutes. Plan an upgrade this quarter; no hard deadline but the perf gains are production-relevant for clusters running many releases.
- CI/CD — Plan: SOURCE_DATE_EPOCH support for reproducible chart archives and expanded GnuPG keybox provenance support are concrete supply-chain hardening gains; also fixes a template regression routing messages to stderr. Schedule adoption this quarter to improve build reproducibility and provenance workflows.
- Leader — Skip
- Platform/SRE — Plan: Minor feature release of Helm with k8s API group updated to 0.37.0; schedule an upgrade this quarter to stay current, but no EOL date or forced migration deadline exists.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Beta (pre-GA) native histograms are now enabled by default in Kubernetes 1.37, changing how latency/duration metrics are exposed by core components like the API server; worth evaluating Prometheus and Grafana compatibility before upgrading clusters to 1.37.
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.37 EOL 2027-10-28 · Kubernetes 1.36 EOL 2027-06-28 · pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Learn: Alpha feature behind a feature gate that closes the ‘deferred’ resize gap introduced when in-place pod resizing went GA in 1.35; worth tracking for future VPA and priority-class design, but no cluster change warranted until it graduates.
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.37 EOL 2027-10-28 · pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Learn: Offers design-level thinking on what makes a platform genuinely AI-ready beyond GPU provisioning — useful for shaping how the team evaluates multi-node training infrastructure, but no operational change required today.
- CI/CD — Skip
- Leader — Learn: Relevant context for platform leaders assessing whether their current infrastructure strategy is positioned for distributed AI workloads — shapes strategic thinking but no immediate decision required.
- Platform/SRE — Learn: Five new standardized Node conditions (DrainInProgress, Drained, MaintenancePlanned, MaintenanceInProgress, GracefulNodeShutdownInProgress) give operators a shared Kubernetes-native way to observe drain and maintenance state. Worth evaluating for alerting and automation integration, but the feature is newly introduced (likely alpha) so no production adoption yet.
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.37 EOL 2027-10-28 · deprecation mentioned (no explicit date found)
- Platform/SRE — Learn: Practical DR scenarios covering the gap between having backups and actually recovering — worth reviewing to pressure-test existing runbooks, but no deadline or current-production gap forces immediate action.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: GA availability of full-caching ephemeral OS disks can meaningfully reduce node boot times and eliminate remote-storage I/O for node pools — worth evaluating for VMSS-backed AKS node groups this quarter.
- CI/CD — Skip
- Leader — Skip
- Signals: GA announcement
- Platform/SRE — Plan: GovCloud operators running EKS with cert-manager and service meshes like Istio or Linkerd should evaluate adopting the Private CA EKS add-on to automate TLS certificate lifecycle; no deadline, but it simplifies a previously manual integration.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Explores patterns for exposing per-tenant GPU utilization metrics without cross-tenant visibility leakage — worth reviewing if you operate shared GPU clusters, but no tooling release or deadline anchors this as actionable now.
- CI/CD — Skip
- Leader — Learn: Addresses the cost-attribution problem for shared GPU infrastructure, relevant if GPU spend is a significant and hard-to-attribute budget line, but this is a design pattern post rather than a strategic decision trigger.
- Platform/SRE — Learn: Kubernetes 1.37 is a GA minor release (EOL 2027-10-28) with no urgent upgrade pressure, and its headline features — gang scheduling, Workload-Aware Preemption, CompositePodGroup — are Beta or newly introduced, capping this at Learn. Worth tracking for teams running AI/ML or large batch workloads on Kubernetes as these APIs stabilize toward GA.
- CI/CD — Skip
- Leader — Learn: The maturing Workload-Aware Scheduling APIs (gang scheduling, hierarchical PodGroups) signal that upstream Kubernetes is building native primitives for AI/ML batch infrastructure that previously required external schedulers — worth incorporating into the AI/ML platform strategy roadmap as these graduate to GA.
- Signals: Kubernetes 1.37 EOL 2027-10-28
- Platform/SRE — Learn: Covers OIDC/IdP integration patterns for on-prem Kubernetes clusters where IAM isn’t bundled — useful for evaluating or hardening auth config, but no deadline or breaking change.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: CNCF graduation signals Karmada is production-vetted; platform teams running multiple clusters or hybrid AI infrastructure should evaluate it as a multi-cluster orchestration candidate this planning cycle.
- CI/CD — Skip
- Leader — Learn: Karmada’s graduation shifts the multi-cluster management landscape—worth tracking as context for build-vs-buy decisions on cross-cluster federation, but no forced decision yet.
- Platform/SRE — Learn: Interesting case study on consolidating AI training and inference onto shared Kubernetes infrastructure, improving accelerator utilization from 35% to 60%+ — useful reference architecture if evaluating GPU workload scheduling.
- CI/CD — Skip
- Leader — Learn: Notable FinOps result: 60%+ reduction in inference cost per million tokens via platform consolidation — worth reviewing as a benchmark when evaluating GPU infrastructure efficiency investments.
- Platform/SRE — Learn: Beta (pre-GA) cap applies, but this is worth evaluating: running kubelet and container runtimes in a Linux user namespace reduces blast radius of node-component CVEs like cr8escape and the 2023 runc procfs bypass; test in non-production clusters before v1.38 targets GA.
- CI/CD — Skip
- Leader — Learn: Rootless-node mode reaching beta signals a maturing Kubernetes security posture option; relevant context for teams in regulated industries evaluating host-isolation requirements, but no adoption decision warranted before GA.
- Signals: Kubernetes 1.37 EOL 2027-10-28 · Kubernetes 1.22 is past EOL (2022-10-28, 1408d ago) · pre-GA (alpha/beta/RC/preview) · CVE-2019-19921 — CISA KEV: not listed, EPSS 0.00 · CVE-2022-0811 — CISA KEV: not listed, EPSS 0.19 · CVE-2023-27561 — CISA KEV: not listed, EPSS 0.00
- Platform/SRE — Plan: DRA Extended Resource support is now stable in 1.37, letting existing extended-resource workloads (e.g. example.com/gpu) route through DRA drivers without ResourceClaims or device plugins — a meaningful adoption path to evaluate this quarter. EOL for 1.37 is 2027-10-28, so no forced upgrade yet; plan an evaluation of DRA adoption for GPU/device workloads in the current planning cycle.
- CI/CD — Skip
- Leader — Learn: DRA reaching GA means GPU and accelerator workloads can now use a unified allocation model without dual plugin stacks — worth understanding for future platform architecture decisions around AI/ML infrastructure, but no toolchain strategy or budget action is required now.
- Signals: Kubernetes 1.37 EOL 2027-10-28
- Platform/SRE — Learn: Practical walkthrough for a common platform hygiene problem — useful for designing or documenting a namespace migration runbook, but no EOL date or active risk requiring immediate action.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Beta (pre-GA) feature that eliminates the need for external add-ons to scale queue consumers and GPU/CPU batch workloads to zero; worth evaluating in non-production clusters now, with a plan to adopt when it reaches GA.
- CI/CD — Skip
- Leader — Learn: The scale-to-zero HPA capability has meaningful cost-reduction potential for GPU and dedicated-CPU workloads, but it is still Beta; track for inclusion in platform cost-optimization standards when it reaches GA.
- Signals: Kubernetes 1.37 EOL 2027-10-28
- Platform/SRE — Plan: Windows Server 2025 is now a supported node OS on AKS, giving teams a clear upgrade target as older Windows Server versions approach end of support. Schedule evaluation of Windows node pool migration this quarter, especially if running 2019 or 2022 nodes — no forced-upgrade date is signaled yet, but the deprecation mention warrants adding it to the roadmap.
- CI/CD — Skip
- Leader — Learn: AKS now supports Windows Server 2025, extending the viability of Windows-based workloads on managed Kubernetes — useful context if the org is evaluating its Windows modernization strategy, but no strategic or cost decision is required now.
- Signals: deprecation mentioned (no explicit date found) · GA announcement
- Platform/SRE — Learn: Interesting integration pattern for teams using KubeVirt and Metal3 together, enabling bare-metal provisioning workflows for VMs. No GA release, deadline, or operational change required; worth evaluating if your platform uses KubeVirt.
- CI/CD — Skip
- Leader — Skip
Learn
archived
Kubernetes v1.37: etcd RangeStream graduates to beta, cuts API server memory on large reads
- Platform/SRE — Learn: RangeStream is still beta in v1.37 (requires etcd v3.7), so not yet production-adoptable, but SREs running clusters with many large objects (e.g., Pods at scale) should track this as a near-term mitigation for API server and etcd OOM risk during cache repopulation.
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.37 EOL 2027-10-28 · etcd 3.7 supported
- Platform/SRE — Plan: CVM node pools on AKS are now GA, enabling sensitive workload isolation at the hardware level; evaluate whether regulated or high-sensitivity workloads in your clusters warrant migrating to CVM node pools this quarter.
- CI/CD — Skip
- Leader — Learn: GA confidential compute on AKS is a new capability relevant to compliance and data-sovereignty positioning, but no immediate strategic decision is required unless the org has active regulated-workload requirements on Azure.
- Signals: GA announcement
- Platform/SRE — Plan: Artifact streaming on AKS+ACR is now GA and can reduce pod startup latency during scale-out events; evaluate enabling it for workloads where image pull time is a bottleneck this quarter.
- CI/CD — Skip
- Leader — Skip
- Signals: GA announcement
- Platform/SRE — Learn: A conceptual overview of Kubernetes observability patterns — useful for shaping how SREs reason about distributed tracing, metrics, and logs across complex workloads, but no new tooling, GA release, or deadline requiring action.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: StorageVersionMigration API (storagemigration.k8s.io/v1) is now stable and enabled by default in Kubernetes 1.37, removing the need for manual migration scripts when promoting or dropping CRD API versions. Plan to incorporate SVM into your CRD lifecycle runbooks when scheduling the upgrade to 1.37 (EOL 2027-10-28).
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.37 EOL 2027-10-28 · GA announcement
- Platform/SRE — Plan: Kubernetes 1.37.0 is a new minor release worth evaluating for adoption this quarter; review the CHANGELOG for API removals or deprecations that may affect running workloads before scheduling an upgrade window.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: A new Istio minor release is always a candidate for upgrade planning — review the full release notes for breaking changes, API removals, or deprecations before scheduling a mesh upgrade this quarter.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: Pod Certificates and Cluster Trust Bundles reaching GA in Kubernetes 1.37 introduces native X.509/mTLS workload identity as an alternative to service account JWTs; evaluate adopting cluster trust bundles and pod certificate issuance this quarter for services requiring mTLS.
- CI/CD — Skip
- Leader — Learn: Native X.509 workload identity baked into Kubernetes core shifts how orgs can approach service-to-service auth without a service mesh; worth tracking as input to future golden-path and identity-standards decisions.
- Signals: Kubernetes 1.37 EOL 2027-10-28
- Platform/SRE — Learn: Covers a real incident pattern — GPU pods pending during traffic spikes — and predictive scaling approaches; worth reading to inform GPU cluster design, but no GA tool, deadline, or breaking change anchors an action now.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: The metrics.k8s.io/v1 API is functionally identical to v1beta1 — no field changes, no behavioral differences. Worth noting when planning a v1.37 upgrade so any hardcoded v1beta1 API paths in tooling or manifests get updated, but no v1beta1 deprecation deadline is announced.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: A conceptual overview of what platform teams need to consider when extending Kubernetes for AI workloads; no concrete tooling changes or deadlines, but useful for shaping future platform strategy around GPU scheduling and resource management.
- CI/CD — Skip
- Leader — Learn: Relevant framing for leaders evaluating whether their current Kubernetes platform strategy needs to extend to AI/ML workload support — useful context for roadmap discussions, but no decision is forced yet.
- Platform/SRE — Plan: This GA capability lets AKS pods authenticate to SMB file shares via workload identity instead of node-level managed identity, improving least-privilege posture. Evaluate replacing existing managed-identity-based Azure Files mounts with workload identity bindings in your next infrastructure review cycle.
- CI/CD — Skip
- Leader — Skip
- Signals: GA announcement
- Platform/SRE — Plan: If you run Mountpoint in EKS or other memory-constrained environments, upgrading to the latest release lets you set explicit memory targets or rely on automatic container-limit detection, preventing the expansion-over-time instability that previously competed with ML or analytics workloads. No deadline, but worth scheduling as a planned upgrade this quarter if Mountpoint is in your stack.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: A new GA Kubernetes minor release with 16 enhancements graduating to Stable and one deprecation/removal is a direct platform concern; audit the removal for any API or feature you currently use and schedule cluster upgrade evaluation this quarter — no forced-upgrade date was found, so Act isn’t warranted yet.
- CI/CD — Skip
- Leader — Learn: Kubernetes v1.37 reflects continued platform maturity but carries no licensing, cost, or vendor-risk angle and no forced-migration deadline; awareness is useful for roadmap conversations, but no leadership decision is pending.
- Signals: deprecation mentioned (no explicit date found)
- Platform/SRE — Learn: Describes a multi-tenant GPU pooling architecture on Kubernetes for concurrent AI workloads; useful design reference if the org is evaluating shared GPU infrastructure, but no GA tooling or deadline makes this actionable today.
- CI/CD — Skip
- Leader — Learn: Offers a mental model for AI infrastructure as a shared organizational capability; relevant if evaluating whether to build a centralized GPU platform versus per-team provisioning.
- Platform/SRE — Plan: GA Bastion-to-AKS tunneling removes the need for a public API server endpoint or VPN for cluster access; evaluate adopting this as the standard private-cluster access pattern in your AKS environments this quarter.
- CI/CD — Skip
- Leader — Skip
- Signals: GA announcement
- Platform/SRE — Plan: This GA release moves AKS packet forwarding into the kernel via eBPF, potentially reducing latency and CPU overhead for networking-heavy workloads; plan evaluation and enablement on AKS clusters running Advanced Container Networking Services this quarter.
- CI/CD — Skip
- Leader — Learn: AKS is expanding its networking performance story with eBPF-based host routing reaching GA — worth noting as a differentiator when evaluating managed Kubernetes options, but no immediate strategic decision required.
- Signals: GA announcement
- Platform/SRE — Plan: This GA feature removes the need for an identity broker when authenticating multiple user populations (employees, contractors, CI/CD systems) to EKS clusters. Evaluate whether your clusters could simplify their auth architecture by replacing any intermediary OIDC broker with direct per-provider associations.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: A solid explainer on liveness, readiness, and startup probes that may refine how you configure them on workloads, but no operational change required today.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: The managed EKS Argo CD capability now accepts argocd-cm ConfigMap settings, including custom health checks for CRDs that can hold sync waves until resources finish provisioning. If your clusters use this managed capability, evaluate adding custom health checks for your Custom Resources this quarter.
- CI/CD — Learn: Custom health check logic for CRDs in EKS-managed Argo CD means sync wave advancement can now be gated on actual resource readiness rather than Argo CD’s default no-op behavior; worth factoring into GitOps deployment design if your org uses this specific managed capability.
- Leader — Skip
- Platform/SRE — Plan: EKS clusters created in 2018 have 10-year CAs now approaching expiry (~2028); audit cluster creation dates and schedule CA rotation this quarter — worker nodes must be replaced and external API clients updated to trust the successor CA before activation, which AWS will not do automatically.
- CI/CD — Learn: Pipelines that connect directly to EKS API servers (kubectl, Helm deploys, kubeconfig-based auth) qualify as external clients under the shared-responsibility model and would need CA trust updates during any rotation; no immediate action required but worth noting when rotation is scheduled by Platform.
- Leader — Skip
- Platform/SRE — Learn: Conceptual framing of multi-plane sovereignty architecture that could inform future platform design decisions, but no actionable change required today and no concrete deadline or migration target.
- CI/CD — Skip
- Leader — Learn: Useful strategic context on sovereignty architecture patterns for leaders weighing data-residency or regulatory requirements, but no vendor decision or cost implication is triggered by this piece.
- Platform/SRE — Learn: Kubeflow’s CNCF graduation signals broader enterprise adoption maturity; worth evaluating if your org runs ML workloads on Kubernetes, but no operational change required today.
- CI/CD — Skip
- Leader — Plan: CNCF graduation marks Kubeflow as a de-facto standard for cloud-native MLOps; evaluate whether to include it in the platform golden path for teams running AI/ML workloads this quarter.
- Platform/SRE — Learn: Analytical benchmark on how CPU throttling from limits degrades throughput and increases cost — worth reviewing when setting resource policies for clusters, but no immediate operational change is required.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Demonstrates a fully automated, immutable-OS-based Kubernetes upgrade pattern using Kairos that could inform how teams redesign their node upgrade strategy; no production action required today.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Interesting case study on how Oxide shaped their Kubernetes integrations around real customer needs; worth reading for bare-metal IDP design patterns, but no operational change required.
- CI/CD — Skip
- Leader — Learn: Useful signal on Oxide as a bare-metal cloud alternative and how its Kubernetes story is maturing; relevant if evaluating build-vs-buy for on-prem infra strategy.
- Platform/SRE — Plan: AKS operators can now collect native control plane metrics (API server, etcd, scheduler) through Managed Prometheus without custom exporters — worth scheduling adoption this quarter to close gaps in cluster observability.
- CI/CD — Skip
- Leader — Skip
- Signals: GA announcement
- Platform/SRE — Plan: New GA capability lets EKS cluster admins tune scheduler, controller manager, and API server parameters — e.g. switching to MostAllocated bin-packing to reduce node count. Review the full parameter list and evaluate whether tuning fits your cluster’s resource-utilization or scaling goals this quarter.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: KYAML is a new style-constrained subset of YAML for Kubernetes manifests introduced by SIG CLI (KEP 5295); no operational change required today, but worth tracking as a future standardization target for IaC manifest authoring.
- CI/CD — Learn: Could influence manifest linting or validation steps in deployment pipelines, but this is a style standard with no pipeline-breaking change or actionable deadline.
- Leader — Skip
- Platform/SRE — Learn: Explores using observable policy as code to guide application behavior on Kubernetes — worth reading for platform teams evaluating OPA/Kyverno patterns, but no concrete operational change required.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Useful framing for platform engineers managing GPU workloads on Kubernetes — DRA changes the device-allocation model relative to legacy device plugins and HAMi. No deprecation date or migration deadline exists in the signals, so no action is required now.
- CI/CD — Skip
- Leader — Learn: If the org is building or standardizing an AI/GPU platform on Kubernetes, this comparison informs the build-vs-adopt decision between DRA and HAMi; no strategic urgency exists without a concrete deadline or licensing change.
- Platform/SRE — Plan: This GA release adds per-token inference cost attribution to OpenCost, directly addressing GPU cost visibility for platform teams running AI workloads on Kubernetes. Evaluate upgrading OpenCost to 1.121.0 this quarter if your clusters host inference workloads.
- CI/CD — Skip
- Leader — Plan: First GA implementation of per-token inference cost tracking in an open CNCF tool is a meaningful FinOps development for orgs with growing GPU spend; evaluate adopting OpenCost 1.121.0 as part of your AI cost attribution strategy this planning cycle.
- Platform/SRE — Learn: K8gb’s CNCF incubation signals growing community maturity for a Kubernetes-native GSLB solution worth evaluating if you need multi-cluster or multi-region traffic distribution, but no GA adoption pressure or deadline exists yet.
- CI/CD — Skip
- Leader — Learn: CNCF incubation indicates the project has met governance and adoption thresholds — worth tracking as a potential open-source alternative to proprietary GSLB solutions in the platform strategy.
- Platform/SRE — Plan: This GA expansion lets platform teams consolidate Kubernetes (ESO), Terraform/OpenTofu, and Vault CLI secrets into a single OpenBao-backed store — worth evaluating this quarter as a replacement for fragmented per-tool secret stores, with no forcing deadline yet.
- CI/CD — Learn: GitLab CI/CD secret support landed in v19.0 already; the new ESO and Terraform integrations are primarily platform-side — no pipeline changes required today, but the unified API surface is worth noting for future supply-chain design.
- Leader — Learn: The consolidated single-store model (one audit trail, one access model across Kubernetes, IaC, and pipelines) is worth tracking as a vendor-consolidation data point when revisiting secrets-toolchain standards, but no pricing or license forcing function exists yet.
- Platform/SRE — Plan: TCPRoute and UDPRoute are now stable in the v1 API, making portable L4 routing viable for production workloads like databases, DNS, and VoIP. If you’re already using experimental Gateway API resources, audit for the new gateway.networking.x-k8s.io API group separation to avoid breakage on upgrade.
- CI/CD — Skip
- Leader — Learn: Gateway API continues maturing as the unified Kubernetes networking standard; L4 GA coverage strengthens the case for standardizing on it as the org’s golden-path ingress model over implementation-specific CRDs.
- Platform/SRE — Learn: Solid explainer on how controller-runtime’s list+watch cache works and why reconcilers read from a local copy rather than hitting kube-apiserver directly — useful for platform engineers writing or reviewing custom controllers to avoid memory and consistency surprises in production.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Illustrates real-world gains from optimizing container image pull pipelines for AI workloads; no operational change required, but worth reviewing the architecture patterns if you run similar GPU/AI workloads on Kubernetes.
- CI/CD — Skip
- Leader — Learn: A concrete benchmark (60x image pull improvement) from a major manufacturer adopting cloud-native for AI/ADAS development; useful context for internal platform investment conversations, but no decision is forced.
- Platform/SRE — Learn: Useful pattern for teams running scale-to-zero workloads where liveness/readiness probes inadvertently prevent genuine idle state; worth evaluating KubeElasti’s ProbeResponse approach when designing or reviewing autoscaling configurations.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Early-stage sandbox project exploring disaggregated hardware composability for Kubernetes; worth tracking as a future architectural direction but nothing to evaluate or adopt yet.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: GA resource placement in Fleet Manager enables centralized policy-driven workload distribution across multiple AKS clusters, worth evaluating this quarter if you operate a multi-cluster Azure environment.
- CI/CD — Skip
- Leader — Learn: Fleet Manager’s GA multi-cluster resource placement matures Azure’s managed Kubernetes offering and may influence build-vs-buy decisions around homegrown multi-cluster orchestration tooling.
- Signals: GA announcement
- Platform/SRE — Learn: Preview feature for Azure Kubernetes Fleet Manager that lets you set a failure threshold before halting fleet-wide update runs — worth watching if you manage multi-cluster AKS fleets, but not yet GA so no action today.
- CI/CD — Skip
- Leader — Skip
- Signals: pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Learn: EKS Provisioned Control Plane clusters now get significantly faster HPA-driven scaling with no configuration changes required — worth knowing if you run large clusters with many HPA objects, but no action needed.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Public preview feature for reducing AKS node startup times on GPU/AI/Windows workloads by pre-baking images; worth evaluating if you run performance-sensitive node pools, but not actionable until GA.
- CI/CD — Skip
- Leader — Skip
- Signals: pre-GA (alpha/beta/RC/preview)
- Platform/SRE — Plan: etcd backs every Kubernetes control plane, and a breaking change in a patch release is unusual — review the v3.7.1 CHANGELOG and upgrade guide before applying this update to any cluster, and validate in a non-production environment first. No hard deadline exists, but the breaking-change flag makes this a planned, careful upgrade rather than routine patching.
- CI/CD — Skip
- Leader — Skip
- Signals: breaking-change flagged
- Platform/SRE — Plan: etcd is the Kubernetes control plane’s backing store — a breaking-change flag in even a patch release means reviewing the upgrade guide before rolling this out to production clusters. No deadline is given, but operators should validate against their environment this quarter before routine patching.
- CI/CD — Skip
- Leader — Skip
- Signals: breaking-change flagged
- Platform/SRE — Plan: etcd backs every Kubernetes control plane, and a breaking-change flag on a patch release is unusual — review the CHANGELOG and upgrade guide before rolling this out to clusters; no hard deadline, but unreviewed breaking changes in a core datastore warrant a planned change window rather than routine rollout.
- CI/CD — Skip
- Leader — Skip
- Signals: breaking-change flagged
- Platform/SRE — Learn: Covers cross-cluster federation patterns for multi-region failover — useful for designing resilient platform architecture, but no GA tooling announcement or deadline makes this actionable today.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Practical post-mortem on how Cilium networking caused GPU underutilization in Kubeflow training jobs — worth reading for anyone operating GPU clusters or eBPF-based CNIs where pod-to-pod latency affects collective communication.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: GDC for VMware 1.35.300-gke.87 (Kubernetes 1.35.3) and 1.34.700-gke.93 (Kubernetes 1.34.7) are available for download; if running 1.34, note its EOL is 2026-10-27, so schedule an upgrade to 1.35 this quarter before that deadline.
- CI/CD — Skip
- Leader — Skip
- Signals: Kubernetes 1.35 EOL 2027-02-28 · Kubernetes 1.34 EOL 2026-10-27
- Platform/SRE — Learn: Confidential Containers reaching incubating status signals growing ecosystem support for hardware-based memory isolation in Kubernetes workloads; worth tracking for future adoption in high-compliance environments, but not yet GA.
- CI/CD — Skip
- Leader — Learn: CNCF incubation indicates the confidential computing pattern is maturing toward a viable standard; relevant to strategy planning for regulated industries, but no adoption decision is warranted yet.
- Platform/SRE — Learn: Survey data confirms Kubernetes as the dominant platform for GenAI workloads; useful for validating architectural direction but no operational change required.
- CI/CD — Skip
- Leader — Learn: CNCF survey finding that 66% of GenAI-hosting orgs run on Kubernetes is useful context for platform strategy and build-vs-buy conversations around AI infrastructure.
- Platform/SRE — Learn: Covers architectural patterns for running databases across multiple Kubernetes clusters with regional-failure resilience — worth reading to inform future stateful workload design, but no GA tool announcement or deadline requiring action.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Interesting technique for testing Kyverno policies by simulating production context; useful for validating policy behavior without live cluster risk, but no operational change required.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Binary Authorization now GA-supports post-quantum cryptography keys (ML-DSA-65/Dilithium3) for attestors — worth noting for future supply-chain hardening plans, but no current deadline or forced migration. Anthos patch releases carry no noted security fixes per the summary.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Koreo offers a unified config-management and resource-orchestration layer on Kubernetes, positioned as an alternative to Helm/Kustomize complexity and Crossplane limitations. Too early and community-unproven to plan adoption, but worth tracking as the internal-developer-platform space matures.
- CI/CD — Skip
- Leader — Learn: An emerging OSS approach from a consulting firm that reframes Kubernetes configuration and resource orchestration as a programmable controller layer — worth filing as a signal when evaluating IDP strategy or build-vs-buy decisions on tooling like Crossplane or Helm at scale.
- Platform/SRE — Plan: If running Cilium 1.19.x, this patch fixes a regression that briefly drops established pod connections during agent restart or upgrade; no deadline or KEV, but the availability impact warrants scheduling an upgrade to 1.19.6 this sprint.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: Two notable bug fixes: a regression that prevents Cilium from starting in kvstore mode with KPR enabled when etcd is behind a Kubernetes service, and incorrect policy denials for L7 load-balanced services on remote identity changes. If running Cilium 1.18.x in either of these configurations, schedule the patch update this sprint.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: A self-hosted alternative to CAST AI for cluster rightsizing and bin-packing consolidation — worth evaluating if cost optimization is on the roadmap, but at 63 stars and no enrichment signals confirming GA stability, treat as an early-stage tool to watch rather than adopt.
- CI/CD — Skip
- Leader — Learn: Signals a maturing open-source alternative to commercial Kubernetes cost-optimization vendors like CAST AI; worth tracking as a build-vs-buy data point for FinOps tooling, but too early-stage to drive a strategic decision today.
- Platform/SRE — Learn: Speculative design essay on a hypothetical K8s 2.0 — useful for shaping long-term mental models on Kubernetes architecture, but no GA release, no deadline, and no operational change required today.
- CI/CD — Skip
- Leader — Learn: Thought-provoking framing on where Kubernetes complexity may drive the ecosystem — worth reading for long-term platform strategy thinking, but no decision or vendor action required.
- Signals: major release (2.0)
- Platform/SRE — Learn: Useful reference for platform engineers evaluating GPU workload patterns on Kubernetes; no forced migration or deadline, but shapes how you’d design node pools and scheduling for LLM inference.
- CI/CD — Skip
- Leader — Learn: Relevant context for build-vs-buy decisions on LLM inference — self-hosting via vLLM vs managed API services — but no concrete strategic decision is forced by this content.
- Platform/SRE — Learn: A practitioner case study on replacing Kubernetes with systemd for simpler workloads — useful context for evaluating when Kubernetes complexity isn’t justified, but no operational change required.
- CI/CD — Skip
- Leader — Learn: Relevant for strategy discussions about Kubernetes adoption scope; provides a concrete counterpoint when evaluating whether all workloads warrant the operational overhead of a cluster.
- Platform/SRE — Learn: A practitioner case study arguing against Kubernetes for certain workloads; useful for calibrating when managed simpler alternatives are a better fit, but no operational change required.
- CI/CD — Skip
- Leader — Learn: A notable opinion piece with strong HN engagement questioning Kubernetes adoption — relevant as a data point when evaluating whether Kubernetes is the right default for the org’s golden path.
- Platform/SRE — Learn: A new project enabling distributed LLM inference natively on Kubernetes—worth evaluating for teams planning AI/ML serving infrastructure, but no GA status confirmed and no enrichment signals to anchor a higher verdict.
- CI/CD — Skip
- Leader — Learn: Signals a maturing ecosystem for running LLM inference on existing Kubernetes infrastructure, relevant to strategy around AI workload hosting; no near-term decision required.
- Platform/SRE — Learn: A GA Kubernetes logging tool (helm/v0.10.1) that merges multi-container logs into a single timeline and now adds ripgrep-backed remote search via a DaemonSet agent; worth evaluating if the team lacks a lightweight log-tail solution between kubectl and a full ELK/Loki stack.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: HAMi’s CNCF incubating status is a maturity signal for GPU sharing and virtualization on Kubernetes — worth evaluating for teams running AI/ML workloads on shared GPU clusters, but no deadline or breaking change makes this actionable today.
- CI/CD — Skip
- Leader — Learn: CNCF backing validates HAMi as a community-governed option for GPU resource sharing — relevant context for leaders building an AI infrastructure strategy, but no licensing, cost, or vendor-risk event requires a decision now.
- Platform/SRE — Learn: Demonstrates a lightweight Kubernetes-native PaaS pattern (DNS/SSL, team management, GitHub integration, Helm chart support) worth evaluating if considering an internal developer platform; no urgency signals and project maturity is unclear for production use.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Flox brings Nix-based reproducible environments into Kubernetes pods, which is an interesting pattern for environment consistency; no GA production deployment case or hard deadline makes this a watch-and-evaluate item.
- CI/CD — Learn: Nix-based environments in Kubernetes could offer reproducible build environments for CI workloads, but no concrete pipeline migration path or GA tooling with deadlines is present.
- Leader — Skip
- Platform/SRE — Learn: Useful pattern for air-gapped or registry-mirrored clusters: configure kubelet to use an internal mirror for the pause/infra image instead of registry.k8s.io. No deadline, but worth evaluating if egress control is a concern.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Surveys the current landscape of on-prem DBaaS patterns and operators (CloudNativePG, Percona, etc.) — useful for shaping how the team exposes database services to app teams, but no deadline or GA feature requiring a change today.
- CI/CD — Skip
- Leader — Learn: Relevant context for platform-as-a-product strategy and build-vs-buy decisions around internal database provisioning, but no licensing change or cost event requiring a decision now.
- Platform/SRE — Plan: ingress-nginx is one of the most widely deployed Kubernetes ingress controllers; its retirement means planning a migration to an alternative (e.g., Envoy Gateway, NGINX Gateway Fabric, or another Gateway API-conformant controller). No forced migration date is confirmed yet, so scope the migration project now before community support winds down.
- CI/CD — Skip
- Leader — Plan: If ingress-nginx is part of the org’s Kubernetes golden path or standard stack, its retirement requires evaluating replacement ingress controllers and updating platform standards; begin that toolchain review this planning cycle before the project loses maintainer support.
- Signals: deprecation mentioned (no explicit date found)
- Platform/SRE — Learn: A practical pattern for restricting Kubernetes egress via Squid proxy — worth evaluating if the team lacks an egress-filtering strategy, but no deadline or GA release anchors this as Plan.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: A conceptual piece on how to reason about Kubernetes; useful for building or refining mental models but no operational change required.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Valuable engineering case study on extreme-scale Kubernetes control plane constraints, scheduler behavior, and etcd limits — useful for informing architecture decisions on large clusters even if most won’t operate at this scale.
- CI/CD — Skip
- Leader — Learn: Illustrates the ceiling of managed Kubernetes scalability on GKE, which informs build-vs-buy decisions for large-scale platform strategies.
- Platform/SRE — Learn: An early-stage open-source project applying agentic AI to Kubernetes SRE workflows is worth evaluating, but no GA signal or production track record exists to justify adoption yet.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: New managed-cluster product layering provisioning, add-ons, and app configs on top of k3s/Hetzner — worth noting as a low-cost Kubernetes option, but no action required and no enrichment signals anchor anything more than awareness.
- CI/CD — Skip
- Leader — Learn: Hetzner-backed k3s as a cost-cutting alternative to EKS/GKE/AKS is a legitimate signal for teams watching cloud spend, but this is a product launch with no pricing data or migration path to evaluate yet.
- Platform/SRE — Learn: A scaling/reliability case study from Databricks on custom Kubernetes load balancing — worth reading for design ideas but no operational change required.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: A solid walkthrough for extending HPA with custom signals (queue depth, connection counts) via a Prometheus-compatible exporter — useful design reference but no operational change required to existing clusters.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: An early-stage open-source project applying agentic AI to Kubernetes SRE workflows; worth watching for future evaluation but no enrichment signals, GA status, or operational urgency to act on now.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Explores the architectural tradeoffs of running each AI agent in its own Pod/ServiceAccount versus a shared runtime on Kubernetes — useful context for platform engineers who may be asked to support AI agent workloads. No action required today.
- CI/CD — Skip
- Leader — Learn: Offers mental-model framing for how AI agent workloads map onto Kubernetes primitives, which could inform a platform strategy for AI/ML infrastructure — but no vendor, licensing, or cost decision is at stake.
- Platform/SRE — Learn: Useful reference for platform teams evaluating Headlamp as a cluster UI, covering auth model differences (kubeconfig vs service-account token) and plugin extensibility. No EOL date for Kubernetes Dashboard is cited, so no urgency—worth a read before the next tooling review cycle.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Useful pattern for operators managing Kubeflow: the plugin surfaces notebook servers, training jobs, and pipelines as first-class resources in Headlamp rather than requiring kubectl fallback. Worth evaluating if the cluster hosts ML workloads, but nothing running today requires a change.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Plan: Flux v2.9.2 fixes a real regression (Kustomization openapi.path URL reconcile failure) introduced in v2.9.1 — worth scheduling an upgrade this sprint if you use that feature. Also note Flux 2.6 passed EOL on 2026-06-30; if still running it, upgrade to 2.7+ now.
- CI/CD — Skip
- Leader — Skip
- Signals: Flux 2.9 supported · Flux 2.7 supported · Flux 2.6 is past EOL (2026-06-30, 13d ago)
- Platform/SRE — Plan: etcd is a critical Kubernetes control-plane component; v3.7.0 carries flagged breaking changes requiring review of the upgrade guide before any cluster upgrade. Plan the migration this quarter — no forced deadline in the signals, but breaking changes mean this needs a scoped project, not a routine bump.
- CI/CD — Skip
- Leader — Skip
- Signals: breaking-change flagged
- Platform/SRE — Plan: etcd is the Kubernetes control-plane backing store, so any new minor release warrants a changelog review for deprecations, API changes, or breaking behavior before scheduling an upgrade cycle; no deadline or CVE signals are present to force earlier action.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: Reflects on the hidden complexity costs of building internal platform abstractions that replicate Kubernetes primitives — useful framing for platform design decisions but no operational change required.
- CI/CD — Skip
- Leader — Learn: Offers strategic perspective on when internal platform layers add complexity rather than value — relevant for leaders evaluating build-vs-buy and IDP investment decisions.
- Platform/SRE — Learn: A desktop GUI for Kubernetes cluster management is worth evaluating as a productivity tool, but no production impact or urgency — assess alongside existing tools like Lens or k9s.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: K3k enables lightweight virtual Kubernetes clusters running inside a host cluster, which is worth evaluating for tenant isolation or dev environment use cases, but it has no GA stability signal in the enrichment data to warrant planning adoption now.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Skip
- CI/CD — Skip
- Leader — Learn: AWS is extending AI-assisted diagnostics across all EMR deployment modes; relevant for leaders evaluating whether AI tooling (including MCP-based agents) changes the skills or support model needed for data platform operations.
- Platform/SRE — Learn: A novelty/educational project showing how far Kubernetes internals can be pushed; no production relevance, but interesting for understanding control-plane architecture.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Learn: An early-stage open-source project offering zero-instrumentation eBPF observability and LLM-driven remediation for Kubernetes is worth evaluating, but no GA signal or enrichment data exists to justify adoption planning yet.
- CI/CD — Skip
- Leader — Skip
- Platform/SRE — Act: ingress-nginx reached end-of-life in March 2026 (now four months past); remaining on it means exposure to unpatched CVEs in a critical ingress path with no upstream fixes coming. Audit clusters for ingress-nginx usage and complete migration to a maintained alternative (Envoy Gateway, Ingress-NGINX from F5, Traefik) immediately.
- CI/CD — Skip
- Leader — Act: The SIG Network ingress-nginx controller is retired, making any org standardized on it subject to growing unpatched CVE exposure with no remediation path; this warrants a brief to leadership and a decision on a replacement ingress standard before the vulnerability surface widens further.
- Signals: deprecation mentioned (no explicit date found)
- Platform/SRE — Plan: etcd is the Kubernetes control-plane datastore, so this GA minor release is directly relevant; evaluate adopting v3.7 this quarter, particularly if large result-set latency or v2store remnants are pain points — no forced-upgrade deadline exists yet.
- CI/CD — Skip
- Leader — Skip
- Signals: etcd 3.7 supported · etcd 3.6 supported