The IPs listed in the question should not be used as the Microsoft-hosted agent allowlist.
For Microsoft-hosted Azure DevOps agents:
- There is not a single fixed IP list to whitelist.
- The IP ranges change over time.
- The correct approach is to allow the IP ranges for all Azure regions mapped to the organization’s geography, using the weekly Azure datacenter JSON file.
- For macOS hosted agents, the IP ranges are different because they run in GitHub’s macOS cloud.
The ranges listed in the question are associated with Azure DevOps service connectivity guidance and are not the complete allowlist for Microsoft-hosted agents. The documentation explicitly states that the Azure DevOps service tag or previously mentioned inbound IP addresses don’t apply to Microsoft-hosted agents.
To get the correct IP ranges:
- In Azure DevOps, open
https://dev.azure.com/<your_organization>/_settings/organizationOverviewand note the organization region. - Find the matching geography for that region.
- Allow the IP ranges for every region in that geography, not only the organization’s home region.
- In the weekly file, use the entries named
AzureCloud.<region>for each region.
For high availability and failover:
- Yes, a hosted agent can run in any region mapped to the organization’s geography.
- Because of that, the effective “failover list” is the set of IP ranges for all regions in that geography.
- If the organization is in the European Union, agents are always served from regions within the EU data residency boundary.
- If using macOS hosted agents, they always run in the US.
If maintaining the full geography-wide allowlist is too broad, the documented alternatives are:
- Microsoft Managed DevOps Pools
- Azure Virtual Machine Scale Set agents
Also note:
- Private connectivity such as ExpressRoute or VPN cannot be used to connect Microsoft-hosted agents to a corporate network.
- Traffic from Microsoft-hosted agents to target systems is over the public network.