Fleet Agent CrashLoop and Outdated RKE Metadata After Rancher v2.11.x Upgrade

Article Number: 000022144

Environment

• SUSE Rancher Prime: v2.11.x
• Fleet: v0.12

Situation

After upgrading Rancher to v2.11.x:

• Downstream k3s clusters remain stuck in an updating state.
• The fleet-agent-0 pod enters a CrashLoopBackOff with the error:

panic: leaseDuration must be greater than renewDeadline

• The cluster creation page displays outdated RKE2/k3s versions.

Cause

• The Fleet crashloop is caused by a Fleet 0.12 bug that breaks the leader election configuration.
• The outdated RKE metadata occurs when rke-metadata-config references an older Rancher version, resulting in outdated RKE2/k3s versions being shown.

Resolution

• Force update Fleet clusters

• Navigate to: Rancher UI >> Continuous Delivery >> Clusters

• Select the affected cluster → Click Force Update
• This will bring the Fleet agent back online and resolve the crashloop issue.

• Update RKE metadata configuration

• Navigate to: Rancher UI >> Global Settings >> Settings >> rke-metadata-config

• Edit the settings and update the version-specific URL to match the current Rancher version → Save