Harness "Delegate is not connected" in CI

By Daniel Zoghalchali·Latchkey

The delegate maintains a heartbeat to the Harness Manager. "Not connected" means the Manager stopped receiving heartbeats, usually because the delegate pod is down, cannot reach Harness, or lacks outbound network.

What this error means

The Delegates page shows the delegate as Disconnected/Not connected, and pipelines fail to assign tasks with "delegate ... is not connected".

Delegate task failed to reach any delegate.
Delegate 'ci-delegate' is not connected.
Last heartbeat: 8 minutes ago.

Common causes

The delegate pod is down or crash-looping

A crashed, OOMKilled, or evicted delegate stops heartbeating and shows disconnected.

The delegate cannot reach the Harness Manager

Blocked outbound HTTPS, a proxy issue, or DNS failure prevents the heartbeat from reaching Harness.

How to fix it

Restore delegate health and connectivity

1. Check the delegate pod status and logs for crashes or OOM.
2. Confirm outbound HTTPS to the Harness Manager URL works.
3. Restart the delegate and confirm it shows Connected.

kubectl -n harness-delegate get pods
kubectl -n harness-delegate logs deploy/ci-delegate

Fix proxy or firewall egress

Allow the delegate outbound access (and proxy settings if used) to the Harness Manager and required endpoints.

How to prevent it

• Run more than one delegate replica for redundancy.
• Alert on delegate heartbeat loss.
• Size delegate resources so it does not OOM under load.

Frequently asked questions

Related guides

References

• Terraform documentation
• GitHub Actions documentation