OpenShift / k8s cluster upgrade: Dynatrace Operator detecting cordoned nodes and automatic Maintenance Windows ?

[gilles_tabary]

Hello all.

We use Dynatrace Managed + Dynatrace OpenShift operator cloud-native full-stack.

When we upgrade / update an OpenShift / K8S cluster, OpenShift nodes (==VM) goes into cordoned state. OpenShift sets up automatically a kind of OpenShift maintenance mode ( =~ alerts silenced) on each node, in turn.

Is there anything in Dynatrace OpenShift operator, which detects this "cordoned " state and subsequently handles Dynatrace Maintenance Window ? I feel there should be. How do people do to set up these maintenance ?

Regards.

Gilles

[gilles_tabary]

Hi.

I did not find doc nor forums entries on that. Not much fortune with Google either. So here is my take on this.

AutoTag ocpmaintenance : Set (and unset) automatically on Kubernetes Nodes « marqued » Cordoned, and related hosts and daemonset workloads

o Entities matching type(KUBERNETES_NODE),kubernetesNodeCordoned(true)
o Entities matching type(HOST),toRelationships.isNodeOfHost(type("KUBERNETES_NODE"),kubernetesNodeCordoned(true))
o Entities matching type(CLOUD_APPLICATION),cloudApplicationDeploymentTypes(KUBERNETES_DAEMON_SET),toRelationships.isInstanceOf(type(CLOUD_APPLICATION_INSTANCE),fromRelationships.runsOn(type("KUBERNETES_NODE"),kubernetesNodeCordoned(true)))

+ Maintenance Window including « ocpmaintenance » tagged entities

o Filter by Host, tagged ocpmaintenance
o Filter by Kubernetes node, tagged ocpmaintenance
o Filter by Kubernetes workload, tagged  ocpmaintenance

Any one would have a feed-back or prior experience ? Any comments ?

Regards.

[gilles_tabary]

I also see this MARKED_FOR_TERMINATION thingy

• How to monitor Host Availability when Azure AVD Hosts Auto Shutdown and UP due to Autoscale set?
  https://community.dynatrace.com/t5/Alerting/How-to-monitor-Host-Availability-when-Azure-AVD-Hosts-Au... 
• https://docs.dynatrace.com/docs/shortlink/event-types-info#marked-for-termination 

Shouldn't this be integrated in the Dynatrace OpenShift operator ? It is !

When an OpenShift node goes unschedulable

• Dynatrace OpenShift operator send a MARKED_FOR_TERMINATION event to Dynatrace OpnShift nodes (=HOST)

2026-06-29 17:39 Kubernetes node cordoned. Node might be drained or terminated.
Entity ​my_ocp_host_worker_123456 (10.x.x.x, 10.x.x.x, 10.x.x.x)
Time​ yesterday, 17:39
Type ​Marked for termination
dt.event.allow_frequent_issue_detection​ false
dt.event.description​ Kubernetes node cordoned. Node might be drained or terminated.
dt.event.group_label​ Marked for termination
dt.event.impact_level​ Infrastructure
dt.event.source​Dynatrace Operator
dt.event.timeout​ 15​

• which will prevent Dynatrace from raising a problem when this entity becomes unavailable for the next 60 minutes
• gotcha : if host shutdown detection is enabled, it will trigger an alert even when this mechanism is activated
• gotcha : related entities like daemonset workload are not "protected" by this mechanisme  : so one would have to invent some Maintenance Windows including these related objects

[gilles_tabary · Accepted solution]

See RFE: OpenShift (Kubernetes k8s) cluster upgrade automatic Maintenance Windows.