Skip to content

Latest commit

 

History

History
31 lines (16 loc) · 1.88 KB

faq.md

File metadata and controls

31 lines (16 loc) · 1.88 KB

FAQ

What does MUO stand for?

managed-upgrade-operator

Does MUO silence alerts?

Yes. All non-critical alerts are silenced during an upgrade. Specific "noisy" critical alerts are also silenced and can be found the maintenance section of the ConfigMap.

How does MUO determine which alerts to silence?

Currently this is a manual process. We are working on dashboards and other metrics to help this become a data driven decision.

Does MUO reserve compute capacity?

Yes, if capacityReservation in the upgradeconfig CR is set to true. MUO creates a new upgrade worker machineset for each availability zone with a size of 1 worker node.

NOTE: spec.capacityReservation is an optional field in the upgradeconfig CR. If this is not defined in the upgradeconfig CR the default value is set to true for OCM provider and false for LOCAL provider.

Does MUO maintain correct instance types for each machine pool?

Yes, if capacityReservation in the upgradeconfig CR is set to true. MUO creates the extra compute based on the found instance types of the current machinesets.

How does MUO handle PodDisruptionBudgets that block node draining?

There is a configurable duration that sets how long MUO should respect a PodDisruptionBudget. Upon reaching this duration MUO will forcefully delete the detected pod. This configuration is performed using the spec.PDBForceDrainTimeout field of the UpgradeConfig CR and is measured in minutes:

What happens to a node that is failing to drain NOT due a PodDisruptionBudget?

MUO will forcefully drain these nodes at this duration that is set by RedHat SRE.