Managed DevOps Pool agent error: Exit Code 2 in azure devops pipelines

Question

Hello community,

Since a day or two my Managed DevOps pools is causing problems: it is trying to get a agent from the pool but it seems like it cannot fetch an agent. I tried many times: one time the pipeline will fail within seconds and the other time a pipeline will fail after 15 minutes. Either way, in the MDP I see provisioned agents AND agents that are ready, but as soon as allocation starts a few seconds after the state will become returned and the pipeline will give exit code 2 and cancel. Some things to mention:

• Last week the agents worked just fine
• I checked quota limit and did not exceed
• I checked image version but it is still the same
• It is in private network and al steps are taken to make the pools work: ingested in vnet with a subnet, firewall rules are set (and as mentioned before last week they did work)

Wanted to add:

• maximum agents did not exceed
• parallel jobs did not exceed
• no error has been shown in de MDP pool overview error code

Behavior I see is: Agent is ready -> Allocated -> Returned

Who can help me with this issue??

Answer 1

Hi @Jurriën de Gelder ,

It looks like your pool is provisioning agents but then immediately returning them (or timing out) so your jobs end with exit code 2. That usually means the VM never fully boots up or can’t reach Azure DevOps to register.

Please try below and confirm:

1. Validate your network connectivity

• Run the MDP connectivity script on a VM in your subnet: https://learn.microsoft.com/azure/devops/managed-devops-pools/configure-networking?view=azure-devops&tabs=azure-portal#validate-endpoint-connectivity
• Make sure it can reach the Azure DevOps control plane, the CDN (download.agent.dev.azure.com), agent-package URLs and storage endpoints.

2. Check firewall/NSG/DNS/UDRs

• Ensure outbound HTTPS (TCP 443) to all required Azure DevOps IPs and domains is allowed: https://learn.microsoft.com/azure/devops/organizations/security/allow-list-ip-url
• Review any custom DNS servers or user-defined routes that might block or redirect traffic.

3. Review pool diagnostics & agent logs

• In the Azure portal, go to your MDP pool’s Diagnostics tab—grab the logs around the failure and look for agent startup errors.
• In your YAML, set variables: system.debug: true so you get verbose agent initialization logs in the pipeline run.

4. Consider pre-provisioned (standby) or stateful agents

• If you have Standby agent mode off, new agents can take up to ~15 minutes to spin up—and the pool times out if they aren’t ready. See delays here: https://learn.microsoft.com/azure/devops/managed-devops-pools/troubleshooting?view=azure-devops#delays-in-pipeline-startup
• Either switch to manual/automatic standby agents or use a stateful pool with a short grace period so agents remain online between jobs.

5. Double-check quotas & image settings

• Verify you haven’t hit the per-SKU core quota: https://learn.microsoft.com/azure/devops/managed-devops-pools/prerequisites?view=azure-devops#review-managed-devops-pools-quotas
• If you’re using a custom image and pre-caching the agent, ensure your sentinel file (.complete) is in place per: https://learn.microsoft.com/azure/devops/managed-devops-pools/configure-images?view=azure-devops

Hope this helps!

---

If the resolution was helpful, kindly take a moment to click on and click on Yes for was this answer helpful. And, if you have any further query do let us know.