An Azure networking service that is used to provision private networks and optionally to connect to on-premises datacenters.
Hello @ Stefano Merotta,
Thank you for reaching out to Microsoft Q&A.
The /32 auto-routes from Private Endpoints are still being injected into the peered VNet even when the PE lives in a non-peered subnet, and the Private Endpoint Network Policy no longer lets you override them with a broader UDR (VNet or subnet-level). You’re back to needing a /32 UDR per endpoint, which is exactly the pre-policy experience.
The current implementation narrows the advertised address spaces for normal traffic, but the PE /32 system routes and the network-policy override logic haven’t fully caught up.
Recommended solution right now:
The cleanest workaround Microsoft suggests for this preview behavior is to use Network Security Groups instead of relying on route overrides:
- Enable NSGs on the relevant subnets in the peered VNet.
- Create rules that explicitly allow or deny traffic to the specific Private Endpoint private IPs (or the subnet they sit in).
This gives you the isolation you were aiming for without having to manage one /32 route per endpoint. It’s the same mitigation Microsoft calls out for other routing limitations in subnet peering while the feature matures.
If NSGs don’t fit your design, the other short-term options are:
- Fall back to full VNet peering + Private Endpoint Network Policies (which works as expected today), or
- Place all your Private Endpoints in a subnet that is part of the subnet peering.
Useful links:
- Configure subnet peering
- Manage network policies for private endpoints (UDR/NSG support)
- Virtual network traffic routing (system routes & PE behavior)
Kindly let us know if the above helps or you need further assistance on this issue.
Please "upvote" if the information helped you. This will help us and others in the community as well.