Google app engine creating instances not in 'default internal' ip ranges?

[Ryan Barker]

I am having issues that some of the instances that are being automatically spun up by google app engine are not able to connect to my database. I have a hunch the issue is that the problem relates to ip addresses outside of the range of ip addresses that should be set up for internal machines.  Unfortunately (or fortunately) the problem resolved itself before I could get the ip address of the machines that could not connect.

1) How can I find the logs of internal ip addresses that are being created?

2) What should the internal firewall ip range be set to? It is currently set to 10.128.0.0/9

Thanks

[Julie (cloud platform support)]

If this issue reoccurs, please report it using our issue tracker and specify when it occurred, a screenshot of any error messages, so we can investigate the issue.

The App Engine instance name is the method of identifying specific logs in Stackdriver logging. There are no specific internal IPs that would be specified in the logs since App Engine typically connects to other services via public IP within our DNS range. You can view Stackdriver logs associated with your instance by selecting Logging under the Cloud Console menu and specifying the instance and time frame. This document goes into detail about viewing logs. 

The 10.128.0.0/9 CIDR block is reserved for the primary IP ranges of automatically created subnets when an auto mode network is created. This document goes over the VPC network and how more subnets can be added. If you are using App Engine Flex, instance tags can be used to modify the firewall rules instead of IPs.