Azure Pipelines Self-Hosted Agent Offline - Job Stuck Queued

By Daniel Zoghalchali·Latchkey

The job is waiting for a self-hosted agent that is offline. The agent service stopped, lost connectivity to Azure DevOps, or its registration token expired - so nothing picks the job up.

What this error means

The job stays in the queue indefinitely. In Project Settings → Agent pools the agent shows Offline (grey). Microsoft-hosted runs are unaffected; only the self-hosted pool is starved.

The job is waiting for an available agent in pool 'Default'.
Agent 'BUILD-AGENT-01' is offline.

Common causes

Agent service stopped or host rebooted

The agent runs as a service; if the host rebooted without the service set to auto-start, or the process was killed, the agent goes offline and stops polling.

Lost connectivity or expired credentials

A network outage between the agent and dev.azure.com, or an expired/ revoked PAT used at registration, drops the agent’s connection.

How to fix it

Restart the agent service

Bring the agent process back online on the host.

# Linux (systemd)
sudo ./svc.sh status
sudo ./svc.sh start
# Windows: restart the "Azure Pipelines Agent" service

Re-check connectivity and re-register if needed

1. Confirm the host can reach https://dev.azure.com over outbound 443.
2. If the PAT expired, reconfigure the agent with a fresh token (./config.sh).
3. Set the service to start on boot so a reboot does not leave it offline.

How to prevent it

• Run agents as auto-start services and monitor their online status.
• Use a long-lived registration method and rotate PATs before expiry.
• Alert on agents going offline so queued jobs do not pile up.

Frequently asked questions

Related guides

References

• GitHub Actions documentation