I have a project setup where I can connect to a local resource through AppEngine Flexible instances launching on a VPC network that is setup with a Cloud VPN connection to my local firewall.
With the release of Serverless VPC for the us-east1 region, I wanted to replace my setup to use AppEngine Standard Gen2 instances vs Flexible for the cost savings. I setup a Serverless VPC for the region/network my AppEngine app is hosted on and my Cloud VPN connection is configured for, updated my app.yaml accordingly, and pushed a new version.
I keep getting timeout errors for the new version that is trying to use Serverless VPC to connect to my local resource.
Some context:
- The VPC Network is named "portal" and setup to "Auto" mode (auto creation of subnets for each region)
- Cloud VPN is setup as a Classic VPN in the "portal" network with Route-based routing in the us-east1 region, connecting to my remote local 192.168.11.0/24 subnet.
- A route exists on the VPC network for destinations 192.168.11.0/24 to use the Cloud VPN I have setup as the next hop (automatically created)
- With the above, AppEngine Flexible deployments on the "portal" network can connect to my local resource as can any other Compute Engine VM on the "portal" network
- I setup the Serverless VPC connector on the us-east1 region with the subnet 10.8.0.0/28
I'm not too clear how Serverless VPC works so I'm not sure how to even begin troubleshooting. When I click on the route rule for the 192.168.11.0/24 destination, I can see the AppEngine Flexible instances listed along with some "serverless-vpc-access" tagged instances that appear to be on a different subnetwork but using 10.8.0.0/28 IPs.
Should this configuration be working? If not, what changes do I need to make in order to support this?