Skip to content

1.2.13

Choose a tag to compare

View all tags
@github-actions github-actions released this 01 Jul 23:22
1.2.13
e2f08f1

1.2.13 (2026-07-01)

This release fixes an issue in the backfill job which caused timestamps to be dropped from metrics. Additionally, there is new functionality to ease maintenance burdens by automatically detecting several values on Oracle Cloud, as well as removing the need for an external prometheus-config-reloader image.

Key Features

  • Oracle Cloud (OCI) Support: The agent now auto-detects Oracle Cloud Infrastructure and reports oracle as the cloud provider, alongside the existing AWS (EKS), Google Cloud (GKE), and Azure (AKS) detection. On OKE the region is detected automatically. Because OCI's metadata service does not expose the numeric account ID CloudZero uses, set cloudAccountId manually on OCI.

  • Per-Component Environment Variables: Each component now accepts its own components.<component>.env list on top of the chart-wide defaults.env. A component's entry overrides the default on name collision, and value: null removes an inherited variable — so you can set something like GOMAXPROCS on a single component without applying it everywhere.

  • Prometheus Config Reloader Shipped In-Image: The prometheus-config-reloader binary is now built from source and shipped inside the agent image, so a default install no longer pulls it as a separate external image — one fewer image to mirror, scan, and maintain in air-gapped or controlled-supply-chain environments.

Bug Fixes

  • Backfill Metadata Timestamps: The backfill job captures labels and annotations for resources the admission webhook does not observe on its own (those that predate the agent, or have not changed since). These were recorded with an incorrect timestamp rather than the observation time, which could cause affected resources to be labeled and allocated incorrectly. The job now records the correct time. Real-time collection through the webhook server was unaffected.

  • Webhook Replica Spreading: The webhook Deployment's pod anti-affinity selected on a label no pod carries, so the default 3 replicas were never spread across nodes despite the chart advertising HA. The rule now matches the real pod label and spreads replicas as intended.

  • Aggregator Affinity Ignored: aggregator.affinity was defined in values but never wired into the aggregator Deployment, so setting it had no effect. It is now honored, matching every other component.

Build and Infrastructure

  • Hardened release-image publishing.
  • Dependency updates.

Upgrade Steps

helm upgrade --install <RELEASE_NAME> cloudzero/cloudzero-agent -n <NAMESPACE> --create-namespace -f configuration.example.yaml --version 1.2.13

If you rely on label- or annotation-based cost allocation, we recommend upgrading promptly so backfilled resource metadata is recorded with accurate timestamps.

Assets 3
Loading