Is Serverless VPC Access connector a single point of failure? - google-cloud-platform

I am wanting to connect my Cloud Run app to Postgres Cloud SQL instance without assigning the instance a public IP. Seems like the only way to do this is with a Serverless VPC Access connector.
The docs indicate that the Serverless VPC Access connector is billed as 1 e2-micro instance per 100Mbps. Does this indicate that the connector is simply a single e2-micro VM? Is there any redundancy/automated-failover configured behind the scenes?
I can't find any SLA for the Serverless VPC Access and am worried that it could be a single point of failure for my app that brings down all DB connections.

The VPC Access Connector is a Compute Engine instance privately managed by Google Cloud. You are billed per 100Mbit of capacity. The instance size can scale up but not back down. Is this a single point of failure, yes but the service will auto recover. Fault tolerance, recovery time and SLA are not published (AFAIK).
Additional information:
The images for the VPC Access Connector instances are from the project serverless-vpc-access-image.
These instances use RFC1918 addresses that cannot overlap your VPCs.
These instances are basically NAT Gateways and require IP forwarding be allowed constraints/compute.vmCanIPForward.

Related

Can I run Cloud Run applications on a private IP (inside dedicated VPC network)?

We have an app running on cloud run and it is authenticated only from API gateway.
But still cloud run has *.run.app public domain associated with it and seems like it can still be security issue for sensitive applications which deal with PII data.
How can we run the cloud run inside private VPC network so that private IP is assigned to it?
Is this a con for cloud run over GKE in terms of private VPC network?
Cloud Run cannot have a "private" IP for your service. In general, Cloud Run will be always have its own *.run.app.
Said that what you can do is to restrict the ingress of the service but you should keep in mind that if you set the service as Private or Private + Load Balancer it will be not reachable by API Gateway but by resources in the VPC.
Of course you can set an Internal Load Balancer + MIG as a proxy + Cloud Run private ingress but this increases the configuration overhead.
I think this will change in the future since there is a Feature Request to support Internal HTTPS Load balancers + Serverless NEGs and with the ingress Internal and Cloud Load Balancing you will have a "private" IP for your service (You can ask access for the preview here).
Answering your last question Is this a con for cloud run over GKE in terms of private VPC network? This is something you should evaluate according to your requirements and in general this particular question is an opinion-based which is off-topic. Consider the facts and choose what is better for you.

GCP: Serverless VPC connector stopped flowing data

I have one VPC Serverless connector which helps Cloud Functions to access the default VPC and then transit to another peered VPC from another project.
The setup was working fine until yesterday.
No change happened.
Now my functions are timing out because the call to internal IPs do not return anything.
I activated the Flow logs for the VPC subnet on which the cloud functions are deployed (us-central1) and they do not show any activity at all, which points me at the Serverless VPC connector.
However it does not have any option to enable log.
I cannot try the network intelligence connectivity test since serverless infra has no IP.
I tried creating another VPC Connector with a different internal IP range, still no flow.
Any idea how I could debug this?

Not able to use Serverless VPC Access for Cloud Run and Compute Engine

I have a Cloud Run service and a Compute Engine VM instance, both are in europe-north1 region.
I would like to connect Cloud Run to Compute Engine VM Instance's internal IP address. For that I tried to create a 'Serverless VPC Access'. When I see the supported regions, there are europe-west[1-3] but not europe-north... And the documentation says that:
In the Region field, select a region for your connector. This must match the region of your serverless service
Does this mean that I cannot use Serverless VPC Access if my services are in europe-north1?
Nevertheless, I created the VPC in europe-west3, thinking that it is the closest one, with suggested IP range: 10.8.0.0/28. However, when I go to CloudRun>service>Edit&Deploy New Revision>Connections tab, I don't get the VPC Connector listed in dropdown box. It has already been 30 mins that I created the connector. Does it take more time to appear?
Europe-north1 isn't a supported region for serverless vpc connector.
If you created a serverless VPC access in europe-west3, it is immediately available for Cloud RUn (or other services). If you don't see it, I think it's because your Cloud Run service isn't in the same region. Only the compliant serverless VPC connectors are shown (and available).

Can not connect between Cloud Run and Compute engine using Internal IP

I have a service which runs on Cloud Run, and a MYSQL, MongoDB databases on Compute Engine. Currently, I'm using public IP for connect between them, I want to use internal IP for improving performance, but i cant find solution for this problem, Please help me some ideas, Thanks.
Now is supported. You can use VPC network connector (Beta):
This feature is in a pre-release state and might change or have
limited support. For more information, see the product launch stages.
This page shows how to use Serverless VPC Access to connect a Cloud
Run (fully managed) service directly to your VPC network, allowing
access to Compute Engine VM instances, Memorystore instances, and any
other resources with an internal IP address.
To use Serverless VPC Access in a Cloud Run (fully managed) service,
you first need to create a Serverless VPC Access connector to handle
communication to your VPC network. After you create the connector, you
set your Cloud Run (fully managed) service configuration to use that
connector.
Here how to create: Creating a Serverless VPC Access connector and here an overview about it: Serverless VPC Access example
According to official documentation Connecting to instances using advanced methods
If you have an isolated instance that doesn't have an external IP
address (such as an instance that is intentionally isolated from
external networks), you can still connect to it by using its internal
IP address on a Google Cloud Virtual Private Cloud (VPC) network
However, if you check the services not yet supported for Cloud Run, you will find:
Virtual Private Cloud Cloud Run (fully managed) cannot connect to VPC
network.
Services not yet supported
You can now do that by running this command upon deployment:
gcloud run deploy SERVICE --image gcr.io/PROJECT_ID/IMAGE --vpc-connector CONNECTOR_NAME
If you already have a Cloud Run deployment, you can update it by running the command:
cloud run services update SERVICE --vpc-connector CONNECTOR_NAME
More information about that here
Connecting from Cloud Run Managed to VPC private addresses is not yet supported.
This feature is in development and is called Serverless VPC Access. You can read more here.
If you have a Compute Engine instance running in the same VPC with a public IP address, you can create an SSH tunnel to connect to private IP addresses through the public instance. This requires creating the tunnel in your own code, which is easy to do.

How do I share a Google Cloud SQL privately outside its region

Not finding any solid answers that fit within the scope of my question.
I have a custom VPC established to allow communication between my SQL server and instance groups. My issues are limited connectivity to the SQL server from instances within the same region as the server itself.
Basically, I created a Cloud SQL instance within us-east region.
When I create a VM Instance within the same region as the SQL instance, I have no issues connecting to its private IP.
mysql -h{PRIVATE_IP} -uroot
However, running this same command from an instance in a different region results in a timeout. Both instances are configured the exact same and within the same VPC network.
I let Google allocate IP address pool for me when I created the IP. Created the private network connection within my custom VPC settings and tried tutorials provided in the Cloud Console documentation itself with no luck.
Any help getting me on the right track would be much appreciated. Thank you.
As documented, if you want to connect Cloud SQL from a Compute Engine instance using private IP, your instance must be in the same region as your Cloud SQL instance.
Keep in mind that your Cloud SQL instances are not created in your VPC network, those are created in a Google internal VPC network that then is peered to your VPC network.
Hope this helps!