Cases8 cases

3 to 7 of 8

1. IL-0118 A global change to Kubernetes namespace labels made the networking layer recompute metadata and overwhelmed the control planes of three of OpenAI's largest GPU clusters

   OpenAI2024

   Symptoms

   From 10:15 am PT on 25 November 2024 a large share of both API and ChatGPT traffic failed with timeouts and 503 errors.

   High error rateHigh latencyControl plane unresponsive

   Trigger

   A global change to Kubernetes namespace labels caused the networking layer to recompute metadata. The change had rolled out to the small clusters without problems.

   Root cause

   Configuration change

   Component

   Kubernetes API server

   Contributing factor

   Excessive blast radiusLimited debug access

   Remediation

   • moved workloads out of the impacted clusters by hand
   • with the cloud provider, scaled up the number of control plane nodes while scaling down the number of data plane nodes, which let the control planes recover

   Action items

   • Lock namespace changes so this incident cannot recur
   • Freeze certain types of infrastructure deploys
   • Allowlist identities to prevent a recurrence of this class of incidents
   • Improve the GPU load balancing logic so traffic is re-routed automatically and more effectively when possible
   • Improve change management for fleet-wide infrastructure changes

   View case
2. IL-0060 Kubernetes 1.24 upgrade removed the master node label that Calico route reflectors selected on, dropping all pod routes and taking Reddit down for 314 minutes

   Reddit2023

   Symptoms

   About two minutes after an engineer started the Kubernetes 1.23 to 1.24 upgrade of the cluster running Reddit's legacy stack, just after 19:00 UTC, the site stopped. The CDN edge request graph, which is kept separate from the cluster, showed traffic by status collapsing. Users saw the error page, a homepage with no posts, or plain errors.

   Service outageHigh error rateDNS resolution failuresMonitoring outagePods stuck in ContainerCreatingControl plane lag

   Trigger

   An in-place upgrade of the legacy cluster from Kubernetes 1.23 to 1.24; in 1.24 the node-role.kubernetes.io/master label is removed even from running clusters.

   Root cause

   Breaking change in an upgrade

   Component

   CalicoKubernetes API serverAdmission webhooketcd

   Contributing factor

   Configuration not in codeUnrepresentative stagingUntested backupMonitoring blind spotSingle point of failure

   Detection

   The site broke about two minutes into the upgrade; an incident was opened immediately and responders were on the call within three minutes, working from CDN edge graphs because the cluster's own metrics were gone.

   Remediation

   • deleted the Open Policy Agent webhook configurations to stop API server timeouts
   • terminated all worker nodes of the cluster
   • restored the control plane from the pre-upgrade backup onto the first control plane node
   • switched off the cluster autoscaler until the control plane was stable
   • generated new TLS certificates so replacement control plane nodes could join etcd
   • brought traffic back in steps from 1% to 100%
   • restarted legacy services that had been stopped by hand

   Action items

   • Codify cluster configuration so every choice has a record, and document in detail what cannot be codified
   • Standardise components across clusters and replace bespoke configurations
   • Build tooling to manage the lifecycle of whole clusters so they can be recreated or replicated
   • Add pre-production clusters that can be tested with confidence
   • Improve the backup restore process and make cluster upgrades safe

   Lessons

   Configuration that exists only in a live cluster, edited by hand through a special tool, can be forgotten, as the route reflector setup was, and then break on an upgrade nobody connects it to. A Kubernetes upgrade cannot be reverted without a restore, so the restore procedure has to be kept current and practised. Consistent, codified clusters are what make upgrades testable; bespoke clusters keep producing outages.

   View case
3. IL-0002 CI builder pods could not get AWS IAM credentials because kubelet fell minutes behind its API rate limit

   Zalando2019

   Symptoms

   Builds on Zalando's in-house Continuous Delivery Platform (CDP), which runs CI/CD on Kubernetes, timed out and failed for almost an hour on 11 June 2019. The Kubernetes builder pods could not get AWS IAM credentials.

   High error rateCredentials unavailableControl plane lag

   Trigger

   An unlucky rescaling of the CDP Kubernetes cluster left only one node available for builder pods. Builder pods were created and deleted on that node so quickly that kubelet fell behind.

   Root cause

   Capacity exhaustion

   Component

   kubeletCredential broker for podsKubernetes API serverCI system

   Contributing factor

   Unreviewed defaultsSingle point of failure

   Remediation

   • Scaled the CDP cluster up by hand so builder pods spread over more than one node

   Action items

   • Raise the kubelet QPS limit from the default 5 to 50 in the global cluster configuration

   View case
4. IL-0075 Kubernetes API servers pegged on CPU when the cluster grew to about 290 nodes, traced to Datadog agents polling the API

   Civis Analytics2019

   Symptoms

   After Civis released its new script infrastructure on Kubernetes internally in September 2018, developers reported that kubectl commands were slow, and users intermittently could not start notebooks and services.

   Control plane lagControl plane unresponsiveHigh latencyHigh error rateDNS resolution failures

   Trigger

   Migrating clients' Python, R and container scripts from the in-house Bocce system to Kubernetes (the project called Croquet) pushed the cluster to about 290 EC2 instances under a large script load.

   Root cause

   Capacity exhaustion

   Component

   Kubernetes API serverTelemetry agentetcd

   Contributing factor

   Unknown limitUnverified assumptionMonitoring blind spotNo capacity planning

   Detection

   Developer reports of slow kubectl commands and user reports of notebooks and services failing to start, then high CPU on the master nodes.

   Remediation

   • moved internal users back to the old Bocce infrastructure
   • resized the three master nodes from m4.large to m4.4xlarge
   • changed the Datadog setup so per-node agents no longer polled the API server for service metadata
   • raised max-requests-inflight for non-mutating requests from 400 to 1200
   • grew the etcd EBS volumes from 20 to 100 GB (100 to 300 IOPS)
   • switched client scripts from the ClusterFirst DNS policy to Default

   Action items

   • Upgrade etcd from 2.2.1 to 3.x once Kops supports a safe migration
   • Solve the in-cluster DNS timeouts so client scripts can talk to other workloads in the cluster

   Lessons

   Kubernetes can reach thousands of nodes only after tuning, and third-party DaemonSets are a common cause of control plane load: an agent on every node that polls or watches the API server multiplies with the cluster size. Datadog, kube2iam and the fluentd metadata filter all showed this. A small configuration change often fixes it, but finding it took load testing and the API server access logs.

   View case
5. IL-0031 Kubernetes API server OOM-killed, leaving the Skipper ingress without routes; the talk points to a misconfigured CronJob manifest

   Zalando2018

   Symptoms

   Customers were affected, and Zalando's IAM service answered requests with HTTP 404. The talk shows this on slides titled "Customer impact" and "IAM returning 404", followed by a slide on the number of pods in the cluster.

   High error rateControl plane unresponsiveOOMKilled

   Trigger

   A CronJob (apiVersion batch/v2alpha1, schedule "*/15 9-19 * * Mon-Fri") whose concurrencyPolicy: Forbid and job history limits sat inside the pod template spec instead of the CronJob spec. The talk shows it right after the slides on the number of pods and the API server being OOM-killed, and the lessons add a quota on the number of pods.

   Root cause

   Capacity exhaustion

   Component

   Kubernetes API serverIngress controller

   Contributing factor

   Single point of failureExcessive blast radius

   Remediation

   • fixed the CronJob manifest: moved concurrencyPolicy: Forbid and the job history limits to the CronJob spec, added activeDeadlineSeconds: 120 and changed the schedule to "7 8-18 * * Mon-Fri"

   Action items

   • Make the ingress stay healthy during API server problems
   • Make the ingress keep the last known set of routes
   • Use a ResourceQuota on the number of pods (the example sets pods: "1500")

   Lessons

   An AWS ALB routes traffic to all hosts when all hosts report unhealthy, so the ingress should stay healthy during API server problems. An ingress controller that reads routes from the API server should keep its last known set of routes. Use a quota for the number of pods.

   View case

By root cause4