This topic describes how the system maintains traffic continuity and manages configuration updates during periods of ULB pod unavailability. It outlines how existing traffic remains unaffected while the datapath is active, and explains the reconciliation process that ensures configuration changes are applied once the ULB pods return to service.
If the ULB pods are stopped or temporarily unavailable, existing live traffic is not impacted. Traffic continues to be forwarded using the datapath state that was already programmed before the ULB pods went down.
However, configuration changes made while the ULB pods are unavailable are not applied immediately. This includes starting new application pods, adding, updating, or deleting LoadBalancer services, and adding, updating, or deleting ULB CLI configuration. These changes are reconciled and applied after the ULB pods are running again.