Skip to main content
Reference for the ankra-agent chart installed from oci://registry.ankra.cloud/ankra/ankra-agent. For installation and architecture, see Cluster Agent.
Copy the exact install command from Ankra: it pins the agent version and carries your cluster’s token, and when the cluster has pipeline settings stored, it carries those as further --set flags. Add --set flags for the values below to that command.
  • A cluster you are connecting: Infrastructure → Clusters → Add a cluster → Connect an existing cluster, tab Install the agent. Name the cluster, select Get the install command, then Show the install command.
  • A cluster already in Ankra: Infrastructure → Clusters → <your cluster> → Settings → Upgrades and maintenance, section Agent install command, Create install command.
Create install command makes a new agent token and replaces the previous one, so every install command created before it stops working. The command is shown once: leaving the page clears it.
Both need the agents.manage permission, because the command contains the agent’s token. Without it the connect page reads Your role cannot show the install command.

Values across upgrades

When Ankra upgrades the agent, the agent reads the values you set on its release and passes them to the new release, so a value you set with --set survives the upgrade. Five values are Ankra’s and are not carried: config.ankra_url, config.token and the pipeline values ci_worker_count and ci_storage_class come from Ankra, and image.tag is dropped so the image follows the chart version. The CPU and memory values are carried separately, see Set through Ankra. If the release’s values cannot be read, the agent refuses the upgrade rather than reset your values to the chart defaults. The refusal starts Refused the cluster agent upgrade: the current values of the ankra-agent release could not be read, gives the reason, and ends upgrading without them would reset the customer's chart values to the defaults.

Connection

Workers and Performance

Pipelines

These values shape the Ankra Pipelines steps the agent runs.

Set through Ankra

ci_worker_count, ci_storage_class and the agent’s CPU and memory requests and limits are values Ankra holds per cluster. Every install or upgrade command Ankra generates for the cluster carries them, and a value Ankra holds wins over the one the running Deployment carries. Set them with the CLI:
Read them back with ankra cluster agent resources get and ankra cluster agent ci get. Both set commands say whether the agent took the values right away or takes them at its next upgrade. See ankra cluster agent resources and ankra cluster agent ci.

Kubernetes Watch

Streaming of live resource state to the platform. Only override when you need to tune event volume.

Hosted Logs

The agent can follow the logs of the cluster’s running containers through the API server, up to logs_ship.max_containers at once, and send the lines to Ankra’s hosted log store. Whether it does is decided per cluster in Ankra, and the switch is off until someone turns it on; see Hosted Logs. These values are the local limits.

Image

Security

The chart enforces non-root defaults matching the image’s ankra user (UID/GID 1000) with a read-only root filesystem. Don’t lower these without a rebuilt image.

Scheduling and Service

Metrics

Public IP Reporting

Makes the agent report its public egress IP to Ankra on check-in. Disabled by default; enabling it makes the agent perform an outbound HTTP lookup to public_ip.lookup_url. Leave it disabled for air-gapped clusters.

Diagnostics

Advanced Environment Tuning

Operator-tunable knobs go in extra_env:
See Cluster Agent - Advanced tuning for when to change these.