Cases6 cases

1 to 5 of 6

1. IL-0133 A storm shut down cooling at Nebius's us-central1 data center, and a disk hot-plug bug kept Managed Kubernetes control planes down for hours after power returned

   Nebius2026

   Symptoms

   On 19 August 2026 the first equipment-level signs in Nebius's us-central1 region appeared around 09:40 UTC: the InfiniBand fabric degraded and GPU over-temperature alerts began to accumulate. Customer reports followed shortly after 10:00. At 10:15 on-call engineers were paged on critical component temperature alerts, and the incident was declared at 10:21.

   Control plane unresponsiveService outageMonitoring outage

   Trigger

   A storm-related event at the data center building took the building management system offline and shut down the chilled-water cooling loop.

   Root cause

   Dependency failure

   Component

   Data center power and coolingCloud block storageetcdInfiniBand fabric

   Contributing factor

   Single point of failureMonitoring blind spotNo alert on limitNo runbookOverlapping failures

   Detection

   No facility-level alert fired, because the storm also disabled the building management system that monitors the facility. Engineers acted on component over-temperature alerts that paged on-call at 10:15 UTC, after InfiniBand degradation, GPU over-temperature alerts and customer reports; the temperature rise from about 09:15 was visible on dashboards but had no alert.

   Remediation

   • had the facility operator switch cooling to manual bypass mode and opened rack rear doors to vent heat
   • re-energised tripped racks, control-plane racks first
   • restored the regional infrastructure control plane by manual intervention
   • restarted block-storage host components to clear stale disk attachment state
   • raised the VM recovery rate limits by hand for the region-scale recovery
   • started the affected Managed Kubernetes control-plane instances by hand
   • applied a controlled stop and start to control-plane instances whose data disk had been left unattached
   • identified and started 104 customer VMs that the incident had stopped

   Action items

   • Complete a joint resilience review with the facility operator covering redundancy of facility-management and monitoring systems, cooling that keeps running when those fail, electrical protection of cooling equipment and on-site response
   • Add facility-level environmental alerting on cooling plant state and room temperature trend that pages on-call independently of the building management system
   • Set up a documented emergency escalation path with the facility operator, with an on-call contact tree and guaranteed response times
   • Define an emergency heat-load reduction procedure with criteria, ownership and mechanics for shutting down or degrading workloads during a cooling loss
   • Establish a dependency-ordered region cold-start procedure with clear ownership for each service layer
   • Run regular region-scale recovery drills that cover VMs, storage attachments and Managed Kubernetes clusters
   • Fix the disk-attachment bug that extended the Managed Kubernetes outage and deploy the fix to all production regions
   • Improve the recovery policies Compute users can configure, and move all Managed Kubernetes control-plane nodes to the improved policy
   • Automate reconciliation of stale storage attachment state after abrupt host failures
   • Remove in-region dependencies from monitoring startup and recovery, and extend cross-region monitoring
   • Make regional health checks used for global traffic management reflect real service availability

   Lessons

   Facility monitoring that runs on the same system it watches goes silent in exactly the failure it should report, and with no alert on the room temperature trend there was no practical window to shed heat load in a controlled way. A cold start of a whole region exercised recovery paths built for isolated failures: single-attempt VM recovery, conservative rate limits and stale disk attachments all needed manual work. The longest part of the outage was restoring customer resources, Managed Kubernetes control planes above all, so Nebius is extending its recovery drills to that layer.

   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-0458 Toolforge Kubernetes API server failed to start after a restart, logging "etcd cluster is unavailable or misconfigured"

   Wikimedia Foundation2019

   Symptoms

   On 10 September 2019 the Kubernetes API server of Wikimedia's Toolforge cluster did not come back after the maintain-kubeusers process restarted it at 18:30. To monitoring and to users the whole cluster looked down: Toolforge users could not launch tools with the webservice command on the Kubernetes backend or use kubectl, and an Icinga check against the "All k8s nodes healthy" endpoint started alerting at 18:45.

   Control plane unresponsiveService outageTLS certificate errors

   Trigger

   At 18:30 the maintain-kubeusers process restarted the Kubernetes API server, which forced it to reconnect to etcd with a CA certificate that Puppet changes the night before had left mismatched with the etcd servers.

   Root cause

   Configuration change

   Component

   Kubernetes API serveretcdConfiguration management tool

   Contributing factor

   Change with a delayed effectHarmful manual actionNo runbook

   Detection

   Automated monitoring: an Icinga alert on the toolschecker check http://checker.tools.wmflabs.org/k8s/nodes/ready, about "All k8s nodes healthy", fired at 18:45.

   Remediation

   • Rebooted the Kubernetes master and the etcd cluster, which did not help
   • Took a backup of etcd and reduced the etcd cluster to one node to rule out raft election flaps
   • Brought etcd back to full capacity through a full disaster recovery from the backup
   • Swapped the API server's CA certificate for the one from the etcd servers, which restored services until Puppet reverted it
   • Placed the etcd Puppet CA certificate in /var/lib/puppet/client/ssl/certs/ca.pem so Puppet kept the right CA

   Action items

   • Document the etcd cluster processes
   • Upgrade Toolforge Kubernetes to improve etcd and Kubernetes error reporting and stop needing an API server restart to add a tool
   • Make kube2proxy more resilient to API server failures
   • Audit the Puppet CA certificates so the tools project has no more than one
   • Make the Kubernetes control plane highly available

   Lessons

   Components that often misbehave, such as etcd, drew the responders to the wrong fix, and working on etcd instead of the CA certificates made the outage worse. Puppet-managed certificates are risky, and very old Kubernetes versions log errors that are hard to read. The failure was hard to find because it surfaced at an API server restart rather than at the Puppet change, because the etcd cluster had long-standing undocumented quirks, and because no disaster recovery or node replacement procedure was written down for etcd.

   View case
4. IL-0059 Single-master Kubernetes control plane kept dying because the kubelet restarted a healthy etcd on a failing kubeadm liveness probe

   NRE Labs2018

   Symptoms

   The daily NRE Labs statistics showed lesson activity collapsing, and kubectl could not reach the cluster: `The connection to the server 1.2.3.4:6443 was refused - did you specify the right host or port?`

   Control plane unresponsiveProbe failuresService outage

   Trigger

   Unknown. The liveness probe that kubeadm writes into the etcd static pod manifest (an `etcdctl get foo` against 127.0.0.1:2379, failureThreshold 8, timeoutSeconds 15) started failing. The author first suspected GCP maintenance the night before, but could not reproduce the probe failure by hand and does not know why it began.

   Root cause

   unknown

   Component

   etcdkubeletKubernetes API server

   Contributing factor

   Single point of failureUnreviewed defaultsHarmful manual actionIgnored early signalMonitoring blind spot

   Detection

   The author saw lesson activity drop in the daily NRE Labs statistics during his morning routine.

   Remediation

   • removed the liveness probe section from the kubeadm-generated etcd static pod manifest and applied it between etcd restarts

   Action items

   • Add monitoring, especially for the single-node master
   • Run a basic "down for maintenance" page in Google Cloud Functions that points to Twitter for updates, to redirect to during outages

   Lessons

   Because all of the infrastructure is managed as code, the cluster could be rebuilt quickly, but destroying it before finding the cause also destroyed the evidence. Monitoring matters, especially with a single-node master. Read the restart counts in kubectl output, not just the Running status. Remember the fundamentals instead of expecting the issue to be more complicated than it is, and do not let the urgency of an outage push you into hasty decisions. Put up a maintenance page instead of leaving the site failing halfway.

   View case
5. IL-0072 High etcd write latency made the API servers return stale data, and the production cluster flipped between two views of its state

   Nordstrom2017

   Symptoms

   A user reported that pods were starting up and then vanishing. When the cluster operators looked, kubectl get pods and kubectl get nodes returned two different sets of data that alternated back and forth, as if there were two opinions of the cluster's state.

   Stale or inconsistent API readsStale endpoint routingService outage

   Trigger

   etcd write latency rose far above the healthy single-digit milliseconds; the speakers say trouble starts once the database sync time crosses double digits or several hundred milliseconds.

   Root cause

   Software defect

   Component

   etcdKubernetes API serverIngress controllerPersistent volumes and stateful sets

   Contributing factor

   Unreviewed defaults

   Detection

   A user reported pods starting up and vanishing.

   Remediation

   • restored service on a replacement cluster
   • freed persistent volumes from the old cluster one by one so they could bind in the new cluster
   • updated DNS records that pointed at the old cluster's load balancer names

   Action items

   • Make the Kubernetes API server request quorum reads from etcd

   Lessons

   etcd is extremely sensitive to write latency; once the database sync time crosses double digits of milliseconds, stale reads or badly degraded performance follow. Without quorum reads, a consistent store can still hand out stale data. Documentation for a fast-moving system changes, so read it again regularly. Moving workloads to a new cluster is hard when they hold persistent volumes or when DNS records point at generated load balancer names.

   View case

By root cause5