Cases96 cases

1 to 5 of 96

1. IL-0102 CoreDNS pods on a production EKS cluster ran out of memory under a burst of new pods, endpoints and nodes, taking down the Buildkite web UI, APIs and job dispatch

   Buildkite2026

   Symptoms

   From 22:47 UTC on 25 August 2026 the Buildkite web interface, REST API, Agent API, job queue and notifications stopped working for every customer. No one could create a build from the UI or the REST API, job dispatch was delayed, and some builds that were already running when the outage began stayed stuck until the services recovered.

   Service outageDNS resolution failuresOOMKilledAutoscaler misbehaviour

   Trigger

   At 22:44 UTC an application deploy created a surge of new pods, which used up the free capacity on the existing nodes and made the EKS cluster scale out. Background worker pools with unusually high maximum replica counts were scaling out at the same time and kept asking for more pods while new nodes were still being provisioned. Together they produced an unusually high rate of change in applications, network endpoints and nodes, and with it a sudden rise in demand on CoreDNS.

   Root cause

   Capacity exhaustion

   Component

   CoreDNSContainer memory limitHorizontal pod autoscalerNode pool

   Contributing factor

   No headroomAutomation amplificationRetry stormMonitoring blind spot

   Detection

   The first status post, at 22:55 UTC, eight minutes after the outage began, said something had gone wrong, and at 23:05 UTC the status page said a sitewide outage was being investigated. CoreDNS query duration had been rising as the cluster grew, but a defect in a key monitoring query and too little monitoring of query volume meant this went unnoticed before the outage.

   Remediation

   • paused further application deploys
   • scaled CoreDNS from three to twelve pods
   • raised each CoreDNS pod's memory limit from 512 MiB to 2 GiB
   • expanded the system node pool that runs CoreDNS

   Action items

   • Increase immediate CoreDNS headroom: twelve pods instead of three, a 2 GiB memory limit instead of 512 MiB, and a larger system node pool
   • Enable EKS-managed CoreDNS autoscaling with a tested minimum replica count and enough capacity to spread the replicas across nodes and availability zones
   • Add direct alerts on CoreDNS query duration, query volume, goroutine growth, memory pressure, OOM restarts and available replicas, and track longer-term trends in capacity planning
   • Find other cluster-critical services that lack tested capacity, direct alerting, spread across failure domains, and either safe autoscaling or documented static headroom, and fix them

   Lessons

   A CoreDNS deployment sized once and left at a fixed size can carry a growing cluster for months and then fail on the first burst of churn it cannot absorb. When query processing slows, pending queries take up memory, the pods hit their limits and restart, and client retries keep the service down after the restart. Server-side DNS query duration and volume need their own alerts: here client-side latency was masked by unrelated application speed-ups, and a broken monitoring query hid the server-side trend. Maximum replica counts raised as a temporary measure remove the cap that normally stops runaway scaling when new nodes are slow to arrive.

   View case
2. 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
3. IL-0438 Builder Work became almost entirely unreachable with 504 errors when the Amazon CloudFront VPC Origins feature failed

   ANDPAD2026

   Symptoms

   At about 17:00 JST on Thursday 16 July 2026, Datadog alerted, and at almost the same time 5xx errors began to rise on the Amazon CloudFront distributions that used VPC Origins connections. Builder Work, ANDPAD's job and career service for the construction industry, became almost entirely unreachable.

   Service outageHigh error rate

   Trigger

   The Amazon CloudFront VPC Origins feature failed; AWS Health later reported a capacity constraint in an internal shared subsystem that manages connections for VPC Origins.

   Root cause

   Cloud provider outage

   Component

   CDN and edge networkCloud load balancerIngress controllerWeb application firewall rules

   Contributing factor

   Single point of failure

   Detection

   Datadog raised an alert, and at about the same time 5xx errors were seen rising on the CloudFront distributions that used VPC Origins.

   Remediation

   • Decided to route traffic temporarily through an internet-facing ALB without CloudFront
   • Recreated the internet-facing Ingress, attached it to the Argo Rollouts Rollout resource and promoted it
   • Created a regional WAF for the ALB with the security team
   • Created a new IngressClass with the ACM certificate, the WAF and access log settings
   • Added Ingresses for normal traffic, the admin screens and redirects, which restored routing

   Action items

   • Return to the CloudFront and VPC Origins layout while keeping a way to switch to internet-facing in code
   • Keep the internet-facing ALB, limited to approved CIDRs through spec.inboundCIDRs in IngressClassParams, so commenting out securityGroupInboundCidrs in values.yaml opens it to 0.0.0.0/0
   • Point the Terraform-managed Route53 A record back to CloudFront and keep the ALB alias commented out in code, so uncommenting and applying switches DNS to the ALB

   Lessons

   VPC Origins keeps the ALB off the internet and simplifies the traffic path, but it makes the service depend on an AWS-managed shared subsystem, and when that fails the service stops even though CloudFront is healthy. ANDPAD kept VPC Origins and instead prepared the escape route ahead of time, accepting the cost of an ALB that always runs, because recreating the Ingress and attaching it to the Rollout had taken so long during this incident.

   View case
4. IL-0512 xpro.mit.edu returned 503 for an hour when its Kubernetes/APISIX origin certificate expired after renewal failed

   MIT Open Learning2026sev1

   Symptoms

   At 3:38 PM EDT on 3 June 2026, alerts fired because both the xPRO production CMS and the main xPRO site at https://xpro.mit.edu were returning 503 errors. Fastly, in front of the site, could not connect to the Kubernetes/APISIX origin, because the origin certificate for xpro-web.odl.mit.edu had expired at 3:29 PM EDT.

   Service outageTLS certificate errors

   Trigger

   The origin certificate xpro-cert for xpro-web.odl.mit.edu, issued on 2026-03-05 with a 90-day validity period, expired at 3:29 PM EDT on 3 June 2026.

   Root cause

   Configuration change

   Component

   cert-managerCDN and edge networkAuthoritative DNS servers and records

   Contributing factor

   No configuration testsMonitoring blind spotChange with a delayed effect

   Detection

   Alerts for the xPRO production CMS and the main xPRO site fired at 3:38 PM EDT on the 503 errors, and a Rootly incident was created at 4:08 PM EDT.

   Remediation

   • Deleted and recreated the stale cert-manager certificate resources so that cert-manager issued a fresh origin certificate

   Action items

   • Remove the backend origin domain from the non-OpenEdX xPRO Fastly TLS subscription, so Fastly manages only xpro.mit.edu
   • Deploy the Pulumi update for the xPRO application in Production
   • Trigger cert-manager renewal of the xPRO certificate after the conflicting Fastly validation CNAME is removed
   • Consider tests or policy checks that keep cert-manager-managed origin domains out of Fastly TLS subscriptions
   • Consider alerting earlier on failed cert-manager CertificateRequests and Orders, so stale ACME failures are found before the certificate expires

   Lessons

   Reading the outage as a failed cert-manager ACME renewal was true but incomplete; the missing part was that Fastly had been set to manage validation for a backend origin domain that belonged to cert-manager. The team concluded that TLS certificates and DNS validation records need guardrails around ownership, and that public Fastly frontend domains and Kubernetes/APISIX origin domains must not be treated as interchangeable service domains in TLS automation.

   View case
5. IL-0237 A Kubernetes patch upgrade evicted every Ray Serve head and worker pod at once, and Lucidworks hosted LLMs went down

   Lucidworks2026sev1

   Symptoms

   From 16:18 UTC on 28 May 2026 the Lucidworks AI hosted LLMs in the us-southcarolina region were unavailable: queries to llama-3-8b-instruct, llama-3v2-3b-instruct and phi-4-multimodal-instruct returned 500 or 429 errors. Search, embedding models, the Lucidworks Platform UI and passthrough LLM calls kept working.

   Service outagePod evictionImage pull failures

   Trigger

   A routine Kubernetes patch upgrade on a cluster in us-southcarolina cycled its node pools one after another in quick succession.

   Root cause

   Faulty deployment

   Component

   Node poolModel inference server

   Contributing factor

   No staged rolloutOverlapping failures

   Remediation

   • Deleted the degraded replacement head node that could not pull container images
   • Waited for the LLM image pulls to finish on the replacement nodes
   • Launched new nodes with the required LLM images and redirected Lucidworks AI traffic to them

   Action items

   • Upgrade clusters that host LLM workloads in sequence, with each node pool fully healthy before the next one is upgraded
   • Investigate image pre-loading, such as pre-baked disks or image streaming, for large ML model images
   • Open a support ticket with the cloud provider about the abnormal Docker pull times
   • Coordinate Kubernetes maintenance windows with LLM service owners so they do not clash with ongoing deployments
   • Build tooling to force a Ray cluster failover to the blue or green state instead of waiting for Ray to settle it

   View case

By root cause11