[Bug]: Deployment fails on agent v25.3.2 due to missing default log_level

[mcantinqc]

Description

After upgrading the agent from 25.2.2 to 25.3.2, the deployment fails because the agent expects a mandatory log level configuration. It seems the Helm chart passes an empty string (or nothing) by default, which is no longer accepted by the new agent version.

Error Logs

Missing mandatory env variables
[2026-01-21T15:36:59Z ERROR cws_deployment] Deployment failed: Failed to parse deployment config: Failed to parse deployment configuration: unknown variant ``, expected one of `trace`, `debug`, `info`, `warning`, `error`

Steps to Reproduce

Use the SentinelOne Helm chart with agent version 25.3.2.
Do not explicitly define configuration.env.agent.log_level in your values.yaml.
Deploy/Upgrade the chart.
Observe the pod crash/failure with the error message above.

Expected Behavior

The Helm chart should provide a default value (e.g., info) for the log level environment variable to ensure backward compatibility and successful deployments without manual intervention.

Actual Behavior

The agent fails to start because it receives an empty variant for the log level configuration.

Workaround

Adding the following configuration to the values.yaml fixes the issue:

configuration:
  env:
    agent:
      log_level: "info"

[oded-s1]

@mcantinqc are you using helm charts version 25.3.2?

[mcantinqc]

@mcantinqc are you using helm charts version 25.3.2?

Yes, the behavior is the helm chart pass empty as log_level, but the agent reject it.

[oded-s1]

@mcantinqc, are you sure the agent version matches the charts version?