Our classic release pipelines intermittently fail with “We stopped hearing from agent Azure Pipelines 1. Verify the agent machine is running and has a healthy network connection…”

Abhinav Dutt 0 Reputation points
2026-10-15T05:28:03.9566667+00:00

During the Power Platform Deploy Package task (Power Platform Build Tools 2.x). That task deploys a Dynamics 365 Finance & Operations package and polls for completion, which takes about 65–75 minutes. On every failed run, the deployment on the target environment completed successfully; only the pipeline lost the agent. The step log for the deploy task was never uploaded, so we have no agent-side output for the failure.User's image

User's image

Azure DevOps
0 comments

1 answer

Sort by: Most helpful
Most helpful Newest Oldest

  1. AI answer

    2026-10-15T05:28:22.44+00:00

    This pattern matches an agent-side timeout or loss of heartbeat during a long-running deployment poll, not a package deployment failure in Finance and Operations.

    The supported fixes from Microsoft guidance are:

    1. Use a self-hosted Windows agent for the deployment job.
      The Power Platform package deploy task requires Windows, and Finance and Operations package deployments routinely exceed 60 minutes.
    2. Increase the job/task timeout to at least 240 minutes.
      Microsoft guidance for this exact deployment scenario states that deployments on Microsoft-hosted agents can time out, and the resolution is to switch to self-hosted Windows agents or increase timeoutInMinutes to 240 or higher.
    3. Check whether the agent stopped sending heartbeats.
      Azure Pipelines agents send a heartbeat once per minute. If the service does not receive a heartbeat for five consecutive minutes, the job is marked as failed. That aligns with runs where the environment deployment finishes successfully but the pipeline still fails because the agent connection was lost before logs were uploaded.
    4. Review resource pressure indicators for the failed run.
      Microsoft guidance for the message “We stopped hearing from agent …” says it can indicate the task exceeded the resources allocated to the agent. Enable verbose logs and inspect resource utilization metrics in the pipeline logs to identify CPU, memory, disk, or network exhaustion.
    5. If using a self-hosted agent, verify agent stability during the 65–75 minute polling window:
      • Turn off automatic updates that could reboot the machine.
      • Turn off sleep settings.
      • If hosted on a VM, avoid live migration or maintenance during the run.
      • Check health metrics for disk, memory, processor, page file, and network.

    For the scenario described, the most direct remediation is:

    • move the Power Platform Deploy Package step to a self-hosted Windows agent, and
    • set timeoutInMinutes: 240 or higher for that job.
    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.