Cases6 cases

1 to 5 of 6

1. IL-0171 Karpenter node auto-repair replaced EKS GPU nodes about 10 minutes after they became Ready, in a loop, because the Bottlerocket NVIDIA AMI lacked libdcgm.so and set AcceleratedHardwareReady to False

   Inflab2025

   Symptoms

   GPU nodes for an AI video upscaling job were created normally and reached Ready, and then about 10 to 11 minutes later Karpenter treated each one as unhealthy and replaced it. Every new node went the same way, so the GPU pods kept restarting and the upscaling work could not make progress.

   Node recreation loopGPU unavailable

   Trigger

   Enabling Karpenter's node auto-repair feature gate (nodeRepair: true) and then running GPU pods on nodes from the Bottlerocket NVIDIA AMI.

   Root cause

   Software defect

   Component

   KarpenterNode operating system imageGPU driver

   Contributing factor

   Automation amplificationUnverified assumption

   Detection

   The team noticed GPU nodes being replaced repeatedly after running GPU pods, then found the DCGMError message in the node events.

   Remediation

   • Disabled Karpenter node auto-repair by setting the nodeRepair feature gate to false

   Lessons

   Automatic node repair acts on every unhealthy node condition, including AcceleratedHardwareReady, so an image defect that only breaks GPU health monitoring is enough to have every GPU node terminated and recreated in a loop. Before turning on auto-repair, check which conditions it watches and whether the node image reports them correctly. Disabling auto-repair stops the loop at once but is only a temporary measure; installing the missing library or changing the AMI address the cause but cost more on an immutable OS such as Bottlerocket.

   View case
2. IL-0147 AKS nodes behind an HTTP proxy failed to provision with CSE exit code 31 on 202508 node images, because a refactoring stopped the proxy from being used when downloading uncached Kubernetes binaries

   Microsoft2025

   Symptoms

   Creating or scaling a node pool on AKS 1.31.8 failed: the VMSS instances never finished provisioning. This started after planned maintenance moved the node image from 202507.* to 202508.*, in a cluster where nodes must send all outbound traffic through an internal HTTP/HTTPS proxy and direct egress is blocked. It happened with both Ubuntu and AzureLinux images.

   Scale-up failureUpgrade stall

   Trigger

   A node image auto-upgrade from 202507.* to 202508.* during planned maintenance, on clusters whose Kubernetes patch version was not pre-cached in the new image.

   Root cause

   Software defect

   Component

   Node operating system imageNode pool

   Contributing factor

   Cache-masked failureAutomation amplification

   Detection

   The customer saw VMSS instance provisioning fail with a vmssCSE error in the VMSS activity logs and reported it as a GitHub issue.

   Remediation

   • The customer pinned the node image to a 202507.* snapshot on clusters running 1.31.8
   • The customer upgraded clusters to AKS 1.31.9 or 1.31.10, whose binaries are cached in the 202508.* images
   • The AKS team hot-patched VHD 20250811 so the proxy configuration is used for uncached binary downloads and made it globally available in public regions on 2025-09-04; it also announced the hotfix for VHD 20250820
   • The AKS team told users to reimage their nodes with az aks nodepool upgrade --node-image-only to pick up the hotfixed image

   Action items

   • Apply the proxy fix in all future VHDs

   Lessons

   A bootstrap path that runs only when an artefact is missing from a cache is easy to break without noticing: here only clusters on Kubernetes versions not pre-cached in the node image ever downloaded binaries, so only they met the proxy regression. Testing node provisioning behind a mandatory egress proxy, on older patch versions as well as the newest, would exercise that path. For clusters with restricted egress, pinning the node image or moving to a Kubernetes version cached in the image were both quick workarounds while the fix was shipped.

   View case
3. IL-0112 The daily node rollover put DFDS's EKS nodes on an AMI later recalled for a kernel issue, causing "PLEG is not healthy" NodeNotReady flaps and pod network timeouts

   DFDS2022

   Symptoms

   On 2022-10-31 a planned upgrade of the vpc-cni EKS add-on, started from the deployment pipeline at 13:45 UTC, timed out at 14:10. Only about 10 of the 45 aws-node pods had upgraded, 5 were stuck in Terminating and the other 30 had not been attempted.

   Node NotReadyUpgrade stallIntermittent timeoutsIntermittent connection failures

   Trigger

   At 07:30 UTC on 2022-10-31 the daily node rollover replaced the existing nodes with nodes built from AMI v20221027, the latest AWS EKS node AMI, released on 2022-10-29 and later recalled because of a kernel issue.

   Root cause

   Software defect

   Component

   Node operating system imageLinux kernelkubeletAWS VPC CNINode launch templateTerraform

   Contributing factor

   No staged rolloutAutomation amplificationOverlapping failuresMonitoring blind spotNo health probes

   Detection

   The team noticed nodes flapping between NodeNotReady and NodeReady ("KubeletNotReady PLEG is not healthy") at 14:30 UTC while watching the stalled vpc-cni upgrade, about seven hours after the rollover. Time to detection was 7 hours. Alerting did not catch the NodeNotReady events or the socket growth.

   Remediation

   • cordoned, drained and deleted 15 stuck nodes one at a time to complete the vpc-cni upgrade
   • changed the Terraform infrastructure-modules so a specific AMI could be set
   • rolled the EC2 Auto Scaling group launch configuration back to the previous known good AMI through the pipeline

   Action items

   • Reach out to teams with suggestions about readiness and liveness probes
   • Create a dashboard for readiness and liveness probe coverage
   • Write a wiki article on readiness and liveness probes and their usefulness
   • Modify infrastructure-modules to allow specifying a particular AMI
   • Schedule a discussion on whether to take the latest AMI immediately and how to test new AMIs before they are deployed
   • Implement alerting on abnormal NodeNotReady events and on increased socket use
   • Create a learning path for Kubernetes within the team
   • Collect logs from nodes centrally
   • Create a Prometheus exporter that retrieves the latest available AMI, with a dashboard and alerts

   Lessons

   When a new node AMI has just been rolled out, upgrading other cluster components on the same day makes it hard to tell which change is causing trouble; here the vpc-cni upgrade on the same day confused matters. Taking the latest AMI automatically with no way to pin one meant a rollback first needed a Terraform change. The team also concluded that it needs better communication paths with application teams during incidents.

   View case
4. IL-0146 New AKS clusters with Azure CNI in Central India lost tunnel connectivity because VMs began getting MAC addresses with a 06 prefix, which exposed a dormant bug in bridge-mode Azure CNI

   Microsoft Azure2020

   Symptoms

   In newly created AKS clusters using Azure CNI in Central India, kubectl exec and kubectl logs could time out, and pod-to-pod or pod-to-node traffic could time out. The cluster diagnostics page showed tunnel exec issues. Clusters using kubenet were not affected.

   Intermittent timeouts

   Trigger

   VMs in Central India started being assigned MAC addresses in a much higher range than before (06 prefix rather than the usual 00).

   Root cause

   Software defect

   Component

   Azure CNINode operating system image

   Contributing factor

   –

   Remediation

   • told owners of affected scale-set clusters to run a node image upgrade, or any Kubernetes upgrade such as a patch version, to get the Azure CNI that does not use a bridge
   • told owners of availability-set clusters to scale out and delete the VMs with a high-range MAC address in bridge mode
   • advised recreating affected clusters that were brand new

   Action items

   • Roll out a hotfix that makes transparent mode the default in all regions

   Lessons

   A change in how the platform assigns MAC addresses can wake a long-dormant bug in the network plugin. Knowing which CNI mode the nodes run (bridge or transparent) and keeping node images current decides whether such a bug reaches you.

   View case
5. IL-0029 Slow user-facing services from CPU throttling on containers that used far less CPU than their limit

   Buffer2020

   Symptoms

   Services on Buffer's kops-managed cluster on AWS (about 60 nodes and 1500 containers) answered more slowly than they should have. The container throttling metric showed that many containers were being throttled whatever their CPU usage was, including containers that never came close to their CPU limit.

   CPU throttlingHigh latencyNode NotReady

   Trigger

   CPU limits on user-facing services, enforced through CFS quota by a Linux kernel with a bug that throttled containers even when they were well under their limit.

   Root cause

   Software defect

   Component

   Container CPU limitLinux kernelNode operating system image

   Contributing factor

   Unpatched known bugUnverified assumption

   Detection

   While reviewing the container throttling metric, the team noticed many containers being throttled even though their CPU usage stayed far below their limits.

   Remediation

   • removed CPU limits from all services directly or indirectly on the users' critical path
   • moved those services to dedicated tainted nodes and gave them matching tolerations
   • set each service's CPU request to its peak usage over a few months plus a margin of more than 20%, and set memory requests the same way

   Action items

   • Move nodes to a kernel that contains the CFS throttling fix (4.19 or later), which for Buffer means a kops version whose host image is Ubuntu 20.04
   • After the kernel upgrade, check whether throttling is fully gone and update the post

   Lessons

   A CPU limit can slow a service down even when the service uses a small part of it, because of how the kernel enforces CFS quota on older kernels; the throttling metric shows it, the CPU usage graph does not. Upgrading the kernel is the safer fix. Removing limits works for latency, but only together with honest CPU and memory requests, isolation of the unbounded services, and monitoring of node usage, since unbounded containers can starve the kubelet.

   View case

By root cause1