K8s Deployment Metrics Missing After UMA Upgrade to Version 25.x
search cancel

K8s Deployment Metrics Missing After UMA Upgrade to Version 25.x

book

Article ID: 445532

calendar_today

Updated On:

Products

DX SaaS

Issue/Introduction

After upgrading the Universal Monitor Agent (UMA) from version 2023.x to 25.11.1.38 or later, Kubernetes (K8s) Deployment metrics may disappear from the metric tree.

Specific error messages in the UMA team logs may include:

  • ERROR ClusterInfoClient:147 - Failed to create secure clusterinfo client/etc/clusterinfo/certs/ca-cert.pem
  • WARN ClusterInfoClient:171 - HttpUrl instance is null, couldn't parse serviceURL, pathInfo.
  • WARN ClusterInfoClient:209 - Error while posting data to clusterinfo service.URL is 'nodes/perfmetrics; exception: Invalid URL port: "null"

Cause

Starting with version 25.x, UMA transitioned to using mTLS (Mutual TLS) for ClusterInfo communication. If the deployment manifest (YAML) or Helm chart is not updated to include the necessary mTLS environment variables and certificate secrets, the agent will fail to communicate with the ClusterInfo service, resulting in missing metrics.

Resolution

To resolve this issue, the UMA deployment manifest must be updated to support mTLS. Ensure the following configurations are present in your uma-team deployment:

1. Environment Variables

Add/update/confirm the following environment variables under the container specification:

  • CAAPM_CLUSTERINFO_MUTUALTLS: "true"
  • CAAPM_CONTAINERINFO_MUTUALTLS: "true"
  • CLUSTERINFO_SERVICE_PORT_HTTPS: "8443"

2. Secrets and Volumes

Ensure that the caaiops-uma-certs secret is provisioned and correctly mounted as a volume in the deployment. The agent requires access to the CA certificate (typically at /etc/clusterinfo/certs/ca-cert.pem) to establish the secure connection.

3. Verification

Once the changes are applied, verify the following in the metric tree:

  • The Deployments subtree should be visible.
  • Pod-level metrics (CPU, Memory, Disk, Network) should show as ACTIVE.
  • Check the UMA team logs to ensure the ClusterInfoClient errors have ceased.