Azure Pipelines "We stopped hearing from agent" Lost Communication

The Azure Pipelines service stopped receiving heartbeats from the agent mid-job, so it concluded the agent was lost and failed the job.

What this error means

A running job fails with "We stopped hearing from agent X. Verify the agent machine is running and has a healthy network connection."

##[error]We stopped hearing from agent agent-2. Verify the agent machine is
running and has a healthy network connection.

Common causes

Network interruption

A transient network drop stopped agent heartbeats.

Agent host crashed or was reclaimed

The VM crashed, was OOM-killed, or a spot instance was reclaimed.

Resource exhaustion

CPU/memory pressure made the agent unresponsive.

How to fix it

Check agent health and connectivity

1. Inspect the agent host for crashes, OOM, or network loss at the failure time.
2. Re-run the pipeline once the agent is healthy.

Harden the agent

1. Add memory and ensure a stable network path to dev.azure.com.

How to prevent it

• If you run on GitHub Actions, self-healing managed runners such as Latchkey auto-retry transient failures, so a lost-agent blip does not fail the job. On Azure DevOps, monitor agent health and network stability.

Related guides