Direct VPC egress with a VPC network

You can enable your Cloud Run service or job to send traffic to a VPC network by using Direct VPC egress with no Serverless VPC Access connector required.

Before you begin

  • If you don't already have a VPC network in your project, create one.

  • If you use Shared VPC, see Connecting to a Shared VPC network.

  • If using a VPC subnet, the subnet must be /24 or larger.

  • To ensure you have enough IP addresses available for use across Cloud Run, note the following conditions:

    • The subnet where you plan to deploy Cloud Run services or revisions must have at least a few hundred IP addresses available.
    • At steady state, if the total number of Cloud Run instances using the subnet is 100 or more, reserve enough IP addresses for at least 4 times (4X) the number of instances. When a revision scales down, note that Cloud Run retains its IP addresses for up to 20 minutes. For example, if you upgrade revisions so that revision 1 scales from 100 instances down to zero while revision 2 scales from zero up to 100, Cloud Run retains the revision 1 IP addresses for up to 20 minutes after scaling down. During the 20-minute window in this case, you must reserve at least 800 IP addresses ((100 + 100) * 4).

Limitations

The following limitations apply to both Cloud Run services and jobs:

  • Cloud Run supports throughput of up to 1 Gbps per individual instance. Exceeding this amount results in performance throttling.
  • A Cloud Run usage quota limits the maximum number of instances you can configure to use Direct VPC egress. The maximum number is configured per Cloud Run revision or job execution. To increase the default limits, see how to increase quotas.

  • Cloud Run services and jobs might experience connection breaks during networking infrastructure maintenance events. We recommend that you use client libraries that can handle occasional connection resets.
The following limitations apply to only Cloud Run jobs and not services:
  • Direct VPC egress for Cloud Run jobs is available in Preview only.
  • To ensure proper job executions, only use Direct VPC egress for jobs that don't require more than 8 concurrent instances and be sure to reserve a minimum of 1,024 IP addresses.

The following items are not supported by Direct VPC egress:

  • VPC Flow Logs doesn't provide the name of the Cloud Run service or revision.
  • VPC Flow Logs are not reported from non-VM resources like Cloud Run or on-premises machines.
  • Firewall Rules Logging
  • Packet Mirroring
  • Network Intelligence Center
  • IPv6 traffic
  • Connectivity Tests
  • Private NAT
  • Network tags or service identity in ingress firewall rules.
  • Firewall rules cannot use Resource Manager tags attached to Cloud Run workloads.
  • Cloud Run jobs that run for more than 1 hour might experience connection breaks. These can occur during maintenance events that migrate the job from one machine to another. The container receives a SIGTSTP signal 10 seconds before the event and a SIGCONT signal after the event. After the container receives the SIGCONT signal, retry the connection.

IP address allocation

To place your Cloud Run service or job on a VPC network, you must specify a network and subnet. Cloud Run allocates IP addresses from your subnet.

IP addresses are ephemeral, so don't create policies based on individual IPs. If you need to create a policy based on IPs, such as in firewall rules, you must use the IP address range of the entire subnet.

To change the network or subnet that your service or job uses, deploy a new service revision or execute a new job task that uses the new network and subnet values.

Scale up

To enable fast scale up in the event of a traffic surge, Cloud Run allocates IP addresses before they're needed.

At any given time, there are likely more allocated IP addresses than the number of instances that exist. To ensure that Cloud Run can obtain enough IP addresses, make sure your subnet has at least a few hundred IP addresses available. If the subnet's total number of instances across all Cloud Run services and jobs grows beyond 100, we recommend having at least 4 times (4X) the total number available. If Cloud Run cannot allocate more IP addresses, it cannot start any more service instances or job tasks until more IP addresses become available. If your IP address space is limited, see Supported IP ranges for more options. For IP allocation efficiency and ease of management, place multiple services or jobs on the same subnet.

Scale down

Even after all services or jobs scale down to zero, Cloud Run reserves some IP addresses from the subnet for up to 20 minutes in case services or jobs must quickly scale up again. Each instance requires one IP address, but Cloud Run reserves a minimum /28 subnet mask at the beginning. Once you exhaust all 16 instances, Cloud Run creates a new subnet.

To delete the subnet, you must first delete or redeploy your Cloud Run services or jobs to stop using the subnet, and then wait 1-2 hours.

Supported IP ranges

Cloud Run supports the following IPv4 ranges for your subnet:

  • RFC 1918 (recommended)
    • 10.0.0.0/8
    • 172.16.0.0/12
    • 192.168.0.0/16
  • RFC 6598
    • 100.64.0.0/10
  • Class E (not recommended with on-premises setups)
    • 240.0.0.0/4

Set up IAM permissions

Ensure that Cloud Run has access to the VPC network by using one of the following methods:

  • Cloud Run Service Agent role: By default, the Cloud Run service agent has the Cloud Run Service Agent role (roles/run.serviceAgent) that contains the necessary permissions.

  • Custom permissions: For more granular control, grant the Cloud Run service agent with the following additional permissions on the project:

    • compute.networks.get
    • compute.subnetworks.get
    • compute.subnetworks.use on the project or the specific subnet
    • compute.addresses.get
    • compute.addresses.list
    • compute.addresses.createInternal
    • compute.addresses.deleteInternal
  • Compute Network User role: If you don't use the default Cloud Run Service Agent role or the custom permissions, grant the Compute Network User role (roles/compute.networkUser) on the Cloud Run Service Agent service account by running the following command:

    gcloud projects add-iam-policy-binding PROJECT_ID \
    --member "serviceAccount:service-PROJECT_NUMBER@serverless-robot-prod.iam.gserviceaccount.com" \
    --role "roles/compute.networkUser"

    Replace the following:

    • PROJECT_ID: the ID of your project.
    • PROJECT_NUMBER: the project number where you deploy your Cloud Run service or job.

Deploy a service

Direct VPC egress allows your Cloud Run service to send traffic to a VPC network without a Serverless VPC Access connector. Network costs scale to zero just like the service itself. You can also add network tags directly on Cloud Run service revisions for more granular network security, such as applying VPC firewall rules.

You can configure Direct VPC egress with a service by using the Google Cloud console, Google Cloud CLI, YAML, or Terraform.

Console

  1. Go to Cloud Run

  2. Click Create Service if you are configuring a new service you are deploying to. If you are configuring and deploying an existing service, click the service, then click Edit and deploy new revision.

  3. If you are configuring a new service, fill out the initial service settings page as needed, then click Container(s), volumes, networking, security to expand the service configuration page.

  4. Click the Networking tab.

  5. Click Connect to a VPC for outbound traffic.

  6. Click Send traffic directly to a VPC.

  7. In the Network field, select the VPC network that you want to send traffic to.

  8. In the Subnet field, select the subnet where your service receives IP addresses from. You can deploy multiple services on the same subnet.

  9. Optional: Enter the names of the network tags that you want to associate with your service or services. Network tags are specified at the revision-level. Each service revision can have different network tags, such as network-tag-2.

  10. For Traffic routing, select one of the following:

    • Route only requests to private IPs to the VPC to send only traffic to internal addresses through the VPC network.
    • Route all traffic to the VPC to send all outbound traffic through the VPC network.
  11. Click Create or Deploy.

  12. To verify that your service is on your VPC network, click the service, then click the Networking tab. The network and subnet are listed in the VPC card.

    You can now send requests from your Cloud Run service to any resource on the VPC network, as allowed by your firewall rules.

gcloud

To deploy a Cloud Run service without a connector from the Google Cloud CLI:

  1. Update gcloud components to the latest version:

    gcloud components update
  2. Ensure that the Compute Engine API is enabled for your project:

    gcloud services enable compute.googleapis.com
    
  3. Deploy your Cloud Run service with the following command:

    gcloud run deploy SERVICE_NAME \
    --image=IMAGE_URL \
    --network=NETWORK \
    --subnet=SUBNET \
    --network-tags=NETWORK_TAG_NAMES \
    --vpc-egress=EGRESS_SETTING \
    --region=REGION

    Replace:

    • SERVICE_NAME with the name of your Cloud Run service.
    • IMAGE_URL with a reference to the container image, for example, us-docker.pkg.dev/cloudrun/container/hello:latest. If you use Artifact Registry, the repository REPO_NAME must already be created. The URL has the shape LOCATION-docker.pkg.dev/PROJECT_ID/REPO_NAME/PATH:TAG
    • NETWORK with the name of your VPC network.
    • SUBNET with the name of your subnet. You can deploy or execute multiple services or jobs on the same subnet.
    • Optional: NETWORK_TAG_NAMES with the comma-separated names of the network tags you want to associate with a service. For services, network tags are specified at the revision-level. Each service revision can have different network tags, such as network-tag-2.
    • EGRESS_SETTING with an egress setting value:
      • all-traffic: Sends all outbound traffic through the VPC network.
      • private-ranges-only: Sends only traffic to internal addresses through the VPC network.
    • REGION with a region for your service.
  4. To verify that your service is on your VPC network, run the following command:

    gcloud run services describe SERVICE_NAME \
    --region=REGION

    Replace:

    • SERVICE_NAME with the name of your service.
    • REGION with the region for your service that you specified in the previous step.

    The output should contain the name of your network, subnet, and egress setting, for example:

    VPC access:
      Network:       default
      Subnet:        subnet
      Egress:        private-ranges-only
    

You can now send requests from your Cloud Run service to any resource on the VPC network, as allowed by your firewall rules.

YAML

  1. If you are creating a new service, skip this step. If you are updating an existing service, download its YAML configuration:

    gcloud run services describe SERVICE --format export > service.yaml
  2. Update the following attributes:

    apiVersion: serving.knative.dev/v1
      kind: Service
      metadata:
        name: SERVICE_NAME
        labels:
          cloud.googleapis.com/location: REGION
      spec:
        template:
          metadata:
            annotations:
              run.googleapis.com/network-interfaces: '[{"network":"NETWORK","subnetwork":"SUBNET","tags":"NETWORK_TAG_NAMES"}]'
              run.googleapis.com/vpc-access-egress: EGRESS_SETTING
          spec:
            containers:
            - image: IMAGE

    Replace:

    • SERVICE_NAME with the name of your Cloud Run service. Service names must be 49 characters or less and must be unique per region and project.
    • REGION with the region for your Cloud Run service, which must match the region of your subnet.
    • NETWORK with the name of your VPC network.
    • SUBNET with the name of your subnet. You can deploy or execute multiple services or jobs on the same subnet.
    • Optional: NETWORK_TAG_NAMES with the names of the network tags you want to associate with a service. For services, network tags are specified at the revision-level. Each service revision can have different network tags, such as network-tag-2.
    • EGRESS_SETTING with an egress setting value:
      • all-traffic: Sends all outbound traffic through the VPC network.
      • private-ranges-only: Sends only traffic to internal addresses through the VPC network.
    • IMAGE with the URL of your service container image.

    You can also specify more configuration, such as environment variables or memory limits.

  3. Create or update the service using the following command:

    gcloud run services replace service.yaml

Terraform

To learn how to apply or remove a Terraform configuration, see Basic Terraform commands.

  1. Add the following to your main.tf file:

    /**
     * Copyright 2024 Google LLC
     *
     * Licensed under the Apache License, Version 2.0 (the "License");
     * you may not use this file except in compliance with the License.
     * You may obtain a copy of the License at
     *
     *      http://www.apache.org/licenses/LICENSE-2.0
     *
     * Unless required by applicable law or agreed to in writing, software
     * distributed under the License is distributed on an "AS IS" BASIS,
     * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
     * See the License for the specific language governing permissions and
     * limitations under the License.
     */
    
    # Example configuration of a Cloud Run service with direct VPC
    
    resource "google_cloud_run_v2_service" "default" {
      name     = "cloudrun-service"
      location = "us-central1"
    
      deletion_protection = false # set to "true" in production
    
      template {
        containers {
          image = "us-docker.pkg.dev/cloudrun/container/hello"
        }
        vpc_access {
          network_interfaces {
            network    = "default"
            subnetwork = "default"
            tags       = ["tag1", "tag2", "tag3"]
          }
        }
      }
    }
    

Optionally, make your service public if you want to allow unauthenticated access to the service.

Create a job

Direct VPC egress allows your Cloud Run job to send traffic to a VPC network without a Serverless VPC Access connector. You can also add network tags directly on Cloud Run jobs for more granular network security, such as applying VPC firewall rules.

You can configure Direct VPC egress with a job by using the Google Cloud console, Google Cloud CLI, or YAML.

Console

  1. Go to Cloud Run

  2. If you are configuring a new job, click the Jobs tab and fill out the initial job settings page as needed. If you are configuring an existing job, click the job, then click Edit.

  3. Click Container, Variables & Secrets, Connections, Security to expand the job properties page.

  4. Click the Connections tab.

  5. Click Connect to a VPC for outbound traffic.

  6. Click Send traffic directly to a VPC.

  7. In the Network field, select the VPC network that you want to send traffic to.

  8. In the Subnet field, select the subnet where your job receives IP addresses from. You can execute multiple jobs on the same subnet.

  9. For Traffic routing, select one of the following:

    • Route only requests to private IPs to the VPC to send only traffic to internal addresses through the VPC network.
    • Route all traffic to the VPC to send all outbound traffic through the VPC network.
  10. Optional: Enter the names of the network tags that you want to associate with your service or services. Network tags are specified at the revision-level. Each service revision can have different network tags, such as network-tag-2.

  11. Optional: Enter the names of the network tags that you want to associate with your job or jobs. For jobs, network tags are specified at the execution level. Each job execution can have different network tags, such as network-tag-2.

  12. Click Create or Update.

  13. To verify that your job is on your VPC network, click the job, then click the Configuration tab. The network and subnet are listed in the VPC card.

    You can now execute your Cloud Run job and send requests from the job to any resource on the VPC network, as allowed by your firewall rules.

gcloud

To create a Cloud Run job without a connector from the Google Cloud CLI:

  1. Update gcloud components to the latest version:

    gcloud components update
  2. Ensure that the Compute Engine API is enabled for your project:

    gcloud services enable compute.googleapis.com
    
  3. Create a Cloud Run job with the following command:

    gcloud run jobs create JOB_NAME \
    --image=IMAGE_URL \
    --network=NETWORK \
    --subnet=SUBNET \
    --network-tags=NETWORK_TAG_NAMES \
    --vpc-egress=EGRESS_SETTING \
    --region=REGION

    Replace:

    • JOB_NAME with the name of your Cloud Run job.
    • IMAGE_URL with a reference to the container image, for example, us-docker.pkg.dev/cloudrun/container/job:latest
    • NETWORK with the name of your VPC network.
    • SUBNET with the name of your subnet. You can deploy or execute multiple services or jobs on the same subnet.
    • Optional: NETWORK_TAG_NAMES with the names of the network tags you want to associate with a job. For jobs, network tags are specified at the execution-level. Each job execution can have different network tags, such as network-tag-2.
    • EGRESS_SETTING with an egress setting value:
      • all-traffic: Sends all outbound traffic through the VPC network.
      • private-ranges-only: Sends only traffic to internal addresses through the VPC network.
    • REGION with a region for your job.
  4. To verify that the job is on your VPC network, run the following command:

    gcloud run jobs describe JOB_NAME \
      --region=REGION
      

    Replace:

    • JOB_NAME with the name of your job.
    • REGION with the region for your job that you specified in the previous step.

    The output should contain the name of your network and subnet, for example:

    VPC network:
      Network:       default
      Subnet:        default
    

You can now execute your Cloud Run job and send requests from the job to any resource on the VPC network, as allowed by your firewall rules.

YAML

  1. If you are creating a new job, skip this step. If you are updating an existing job, download its YAML configuration:

    gcloud run jobs describe JOB_NAME --format export > job.yaml
  2. Update the following attributes:

    apiVersion: run.googleapis.com/v1
      kind: Job
      metadata:
        name: JOB_NAME
        annotations:
          run.googleapis.com/launch-stage: BETA
        labels:
          cloud.googleapis.com/location: REGION
      spec:
        template:
          metadata:
            annotations:
              run.googleapis.com/network-interfaces: '[{"network":"NETWORK","subnetwork":"SUBNET","tags":"NETWORK_TAG_NAMES"}]'
              run.googleapis.com/vpc-access-egress: EGRESS_SETTING
          spec:
            containers:
            - image: IMAGE

    Replace:

    • JOB_NAME with the name of your Cloud Run job. Job names must be 49 characters or less and must be unique per region and project.
    • REGION with the region for your Cloud Run job, which must match the region of your subnet.
    • NETWORK with the name of your VPC network.
    • SUBNET with the name of your subnet. You can deploy or execute multiple services or jobs on the same subnet.
    • Optional: NETWORK_TAG_NAMES with the names of the network tags you want to associate with a job. For jobs, network tags are specified at the execution-level. Each job execution can have different network tags, such as network-tag-2.
    • EGRESS_SETTING with an egress setting value:
      • all-traffic: Sends all outbound traffic through the VPC network.
      • private-ranges-only: Sends only traffic to internal addresses through the VPC network.
    • IMAGE with the URL of your job container image.
  3. Create or update the job using the following command:

    gcloud run jobs replace job.yaml

Restrict access with firewall rules

Restrict access to resources in a VPC network by using VPC firewall rules. Add these restrictions by using one of the following strategies:

  • Create an ingress firewall rule that refers to your service or job by using the subnet's IP range.
  • Create an egress firewall rule that refers to your service or job.

    In the egress firewall rule, refer to your service or job by using the linked service account service identity, the subnet's IP range, or the associated network tags.

Network tags for egress

Add an additional layer of network security by using network tags in egress firewall rules.

Console

To associate network tags with a service or job:

  1. In the Google Cloud console, go to the Cloud Run page.

    Go to Cloud Run

  2. Click the service or job you want to associate network tags with, then click Edit and deploy new revision for services or Edit for jobs.

  3. Click the Networking tab for services, or the Connections tab for jobs.

  4. Ensure that you have selected Connect to a VPC for outbound traffic and Send traffic directly to a VPC.

  5. In the Subnet field, select the subnet where your service receives IP addresses from. You can deploy or execute multiple services or jobs on the same subnet.

  6. In the Network tags field, enter the names of the network tags that you want to associate with your service or job.

  7. Click Deploy or Update.

For services, every service revision can have a different set of network tags because network tags are specified at the revision level. For jobs, a job execution has the same network tags that the job had when the execution of the job was created.

gcloud

To associate network tags with a service or job, use the gcloud run deploy command:

gcloud run deploy SERVICE_JOB_NAME \
    --image=IMAGE_URL \
    --network=NETWORK \
    --subnet=SUBNET \
    --network-tags=NETWORK_TAG_NAMES \
    --region=REGION

Replace the following:

  • SERVICE_JOB_NAME with the name of your service or job.
  • IMAGE_URL with the image URL of the service or job.
  • NETWORK with the name of your VPC network.
  • SUBNET with the name of your subnet. You can deploy or execute multiple services or jobs on the same subnet.
  • NETWORK_TAG_NAMES with the name of your network tag or a comma-separated list of network tags.
  • REGION with the name of your region.

For services, every service revision can have a different set of network tags because network tags are specified at the revision level. For jobs, a job execution has the same network tags that the job had when the execution of the job was created.

Disconnect a service

Console

  • To remove your service from the VPC network:

    1. Go to Cloud Run

    2. Click the service you want to remove, then click Edit and deploy new revision.

    3. Click the Networking tab.

    4. Clear Connect to a VPC for outbound traffic.

    5. Click Deploy.

    6. To verify that your service is no longer on your VPC network, click the Networking tab. The network and subnet are no longer listed in the VPC card.

  • To remove only the network tags while keeping the service connected to the VPC network:

    1. Click the service that contains the network tags you want to remove, then click Edit and deploy new revision.

    2. Click the Networking tab.

    3. Clear the names of the network tags that you no longer want to associate with your service.

    4. Click Deploy.

gcloud

  • To remove your service from the VPC network, run the following command:

    gcloud run services update SERVICE_NAME --region=REGION \
    --clear-network
  • To remove only the network tags while keeping the service connected to the VPC network, run the following command:

    gcloud run services update SERVICE_NAME --region=REGION \
    --clear-network-tags

    Replace the following:

    • SERVICE_NAME: the name of your Cloud Run service.
    • REGION: the region for your Cloud Run service.

YAML

  • To remove your service from the VPC network:

    1. Download the service's YAML configuration:

      gcloud run services describe SERVICE_NAME --format export > service.yaml
    2. Remove the following content from your service.yaml file:

      run.googleapis.com/network-interfaces: '[{"network":"NETWORK","subnetwork":"SUBNET","tags":"NETWORK_TAG_NAMES"}]'

      Where

      • NETWORK: the name of your VPC network.
      • SUBNET: the name of your subnet.
      • Optional: NETWORK_TAG_NAMES: the names of the network tags if you had associated them with a service.
    3. Deploy the service revision by running the following command:

      gcloud run services replace service.yaml
  • To remove only the network tags while keeping the service connected to the VPC network:

    1. Download the service's YAML configuration:

      gcloud run services describe SERVICE_NAME --format export > service.yaml
    2. Remove the tags variable from the content in your service.yaml file, leaving the network and subnetwork variables in place, as shown in the following example:

      run.googleapis.com/network-interfaces: '[{"network":"NETWORK","subnetwork":"SUBNET"}]'

      Where

      • NETWORK: the name of your VPC network.
      • SUBNET: the name of your subnet.
    3. Deploy the service revision by running the following command:

      gcloud run services replace service.yaml

Disconnect a job

Console

  • To remove your job from the VPC network:

    1. Go to Cloud Run

    2. Click the job you want to remove, then click Edit and deploy new revision.

    3. Click the Connections tab.

    4. Clear Connect to a VPC for outbound traffic.

    5. Click Update.

    6. To verify that your job is no longer on your VPC network, click the Configuration tab. The network and subnet are no longer listed in the VPC card.

  • To remove only the network tags while keeping the job connected to the VPC network:

    1. Click the job that contains the network tags you want to remove, then click Edit and deploy new revision.

    2. Click the Connections tab.

    3. Clear the names of the network tags that you no longer want to associate with your job.

    4. Click Update.

gcloud

  • To remove your job from the VPC network, run the following command:

    gcloud run jobs update JOB_NAME --region=REGION \
      --clear-network
      
  • To remove only the network tags while keeping the job connected to the VPC network, run the following command:

    gcloud run jobs update JOB_NAME --region=REGION \
      --clear-network-tags
      

    Replace the following:

    • JOB_NAME: the name of your Cloud Run job.
    • REGION: the region for your Cloud Run job.

YAML

  • To remove your job from the VPC network:

    1. Download the job's YAML configuration:

      gcloud run jobs describe JOB_NAME --format export > job.yaml
    2. Remove the following content from your job.yaml file:

      run.googleapis.com/network-interfaces: '[{"network":"NETWORK","subnetwork":"SUBNET","tags":"NETWORK_TAG_NAMES"}]'

      Replace the following:

      • NETWORK: the name of your VPC network.
      • SUBNET: the name of your subnet.
      • Optional: NETWORK_TAG_NAMES with the names of the network tags if you had associated them with a job.
    3. Update the job by running the following command:

      gcloud run jobs replace job.yaml
  • To remove only the network tags while keeping the job connected to the VPC network:

    1. Download the job's YAML configuration:

      gcloud run jobs describe JOB_NAME --format export > job.yaml
    2. Remove the tags variable from the content in your job.yaml file, leaving the network and subnetwork variables in place, as shown in the following example:

      run.googleapis.com/network-interfaces: '[{"network":"NETWORK","subnetwork":"SUBNET"}]'

      Replace the following:

      • NETWORK: the name of your VPC network.
      • SUBNET: the name of your subnet.
    3. Update the job by running the following command:

      gcloud run jobs replace job.yaml

Troubleshooting

Cannot delete subnet

To delete a subnet, you must first delete or redeploy all resources that use it. If Cloud Run is using a subnet, disconnect the Cloud Run service or job from the VPC network or move it to a different subnet before deleting the subnet.

Direct VPC subnet runs out of IP addresses

If the subnet of the VPC network runs out of IP addresses, it is logged by Cloud Logging. When this occurs, Cloud Run cannot start any more service instances or job tasks until more IP addresses become available.

View allocated IP addresses

To see which IP addresses Cloud Run has allocated, go to the IP addresses page in the Google Cloud console or run the following command from the Google Cloud CLI:

gcloud compute addresses list