Cloud Composer 1 is in the post-maintenance mode. Google does not release any further updates to Cloud Composer 1, including new versions of Airflow, bugfixes, and security updates. We recommend planning migration to Cloud Composer 2.

Create Cloud Composer environments

Cloud Composer 1 | Cloud Composer 2 | Cloud Composer 3

This page explains how to create a Cloud Composer environment.

For more information about environments, see Environment architecture.
For more information about creating an environment with Terraform, see Create environments (Terraform).

Before you begin

Enable the Cloud Composer API. For the full list of services used by Cloud Composer, see Services required by Cloud Composer.
The approximate time to create an environment is 25 minutes.

During the environment creation, you specify a service account. Cloud Composer binds this account to a Kubernetes service account of your environment's cluster.

By default, Cloud Composer environments use the Google-managed default Compute Engine service account. We recommend you to create a user-managed service account that has a role specific for Cloud Composer and use it for your environments.

Note: In the past, Cloud Composer users were able to deploy Cloud Composer applications that authenticated as service accounts, even if they didn't have permission to impersonate service accounts. This legacy behavior still affects some organizations. For more information, see Requiring impersonation permissions when attaching service accounts to resources.

If you create an environment with Terraform, the service account used by Terraform must have a role with the composer.environments.create permission enabled.

For more information about the service account for Terraform, see Google Provider Configuration Reference.

For more information about using Terraform to create a Cloud Composer environment, see Terraform documentation.

For more information about additional parameters, see Terraform Argument Reference.
Private IP: There are specific network and peering requirements to create a Private IP environment. For more information, see Configuring private IP.
Shared VPC: There are specific network requirements to use Shared VPC with Cloud Composer. For information, see Configuring shared VPC.
VPC SC: To deploy Cloud Composer environments inside a security perimeter, see Configuring VPC SC. When used with Cloud Composer, VPC Service Controls have several known limitations.

Step 1. Basic setup

This step creates a Cloud Composer environment with default parameters in the specified location.

Console

In the Google Cloud console, go to the Create environment page.

Go to Create environment
In the Name field, enter a name for your environment.

The name must start with a lowercase letter followed by up to 62 lowercase letters, numbers, or hyphens, and cannot end with a hyphen. The environment name is used to create subcomponents for the environment, so you must provide a name that is also valid as a Cloud Storage bucket name. See Bucket naming guidelines for a list of restrictions.
In the Location drop-down list, choose a location for your environment.

A location is the region where the environment is located.
In the Image version drop-down list, select a Cloud Composer image with the required version of Airflow.

gcloud

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version IMAGE_VERSION

Replace:

ENVIRONMENT_NAME with the name of the environment.

The name must start with a lowercase letter followed by up to 62 lowercase letters, numbers, or hyphens, and cannot end with a hyphen. The environment name is used to create subcomponents for the environment, so you must provide a name that is also valid as a Cloud Storage bucket name. See Bucket naming guidelines for a list of restrictions.
LOCATION with the region for the environment.

A location is the region where the environment is located.
IMAGE_VERSION with the name of a Cloud Composer image.

Example:

gcloud composer environments create example-environment \
    --location us-central1 \
    --image-version composer-2.9.11-airflow-2.9.3

API

Construct an environments.create API request. Specify the configuration in the Environment resource.

{
  "name": "projects/PROJECT_ID/locations/LOCATION/environments/ENVIRONMENT_NAME",
  "config": {
    "softwareConfig": {
      "imageVersion": "IMAGE_VERSION"
    }
  }
}

Replace:

PROJECT_ID with the Project ID.
LOCATION with the region for the environment.

A location is the region where the environment is located.
ENVIRONMENT_NAME with the environment name.

The name must start with a lowercase letter followed by up to 62 lowercase letters, numbers, or hyphens, and cannot end with a hyphen. The environment name is used to create subcomponents for the environment, so you must provide a name that is also valid as a Cloud Storage bucket name. See Bucket naming guidelines for a list of restrictions.
IMAGE_VERSION with the name of a Cloud Composer image.

Example:

// POST https://composer.googleapis.com/v1/{parent=projects/*/locations/*}/environments

{
  "name": "projects/example-project/locations/us-central1/environments/example-environment",
  "config": {
    "softwareConfig": {
      "imageVersion": "composer-2.9.11-airflow-2.9.3"
    }
  }
}

Terraform

To create an environment with default parameters is a specified location, add the following resource block to your Terraform configuration and run terraform apply.

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "ENVIRONMENT_NAME"
  region = "LOCATION"

  config {
    software_config {
      image_version = "IMAGE_VERSION"
    }
  }
}

Replace:

ENVIRONMENT_NAME with the name of the environment.

The name must start with a lowercase letter followed by up to 62 lowercase letters, numbers, or hyphens, and cannot end with a hyphen. The environment name is used to create subcomponents for the environment, so you must provide a name that is also valid as a Cloud Storage bucket name. See Bucket naming guidelines for a list of restrictions.
LOCATION with the region for the environment.

A location is the region where the environment is located.
IMAGE_VERSION with the name of a Cloud Composer image.

Example:

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "example-environment"
  region = "us-central1"

  config {
    software_config {
      image_version = "composer-2.9.11-airflow-2.9.3"
    }
  }
}

Step 2. (Optional) Select a service account for your environment

Cloud Composer binds this service account to the Kubernetes service account of your environment. Nodes in your environment's cluster run as the Kubernetes service account and use the bindings to access resources in your Google Cloud project, such as DAG definition files in your environment's bucket.

By default, Cloud Composer environments use the default Compute Engine service account. We recommend you to set up a user-managed service account for Cloud Composer environments.

You cannot change the service account of your environment later.

Console

On the Create environment page, in the Service account drop-down list, select a service account for your environment.

gcloud

When you create an environment, the --service-account specifies the service account for your environment.

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --service-account "SERVICE_ACCOUNT"

Replace:

SERVICE_ACCOUNT with the service account for your environment.

Example:

gcloud composer environments create example-environment \
    --location us-central1 \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --service-account "example-account@example-project.iam.gserviceaccount.com"

API

When you create an environment, in the Environment > EnvironmentConfig resource, specify a service account for your environment.

{
  "name": "projects/PROJECT_ID/locations/LOCATION/environments/ENVIRONMENT_NAME",
  "config": {
    "nodeConfig": {
      "serviceAccount": "SERVICE_ACCOUNT"
    }
}

Replace:

SERVICE_ACCOUNT with the service account for your environment.

Example:


// POST https://composer.googleapis.com/v1/{parent=projects/*/locations/*}/environments

{
  "name": "projects/example-project/locations/us-central1/environments/example-environment",
  "config": {
    "nodeConfig": {
      "serviceAccount": "example-account@example-project.iam.gserviceaccount.com"
    }
  }
}

Terraform

When you create an environment, use the service_account field in the node_config block.

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "ENVIRONMENT_NAME"
  region = "LOCATION"

  config {
    node_config {
      service_account = "SERVICE_ACCOUNT"
    }
  }
}

Replace:

SERVICE_ACCOUNT with the service account for your environment.

Example:

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "example-environment"
  region = "us-central1"

  config {
    node_config {
      service_account = "example-account@example-project.iam.gserviceaccount.com"
    }
  }
}

Step 3. Grant required permissions to Cloud Composer service account

When you enable Cloud Composer API in your project, the Composer Service Agent account is created in your project. Cloud Composer uses this account to perform operations in your Google Cloud project.

Cloud Composer v2 API Service Agent Extension role provides additional permissions to Cloud Composer Service Agent account. This role is not granted automatically. You must grant it manually.

Console

When you create an environment in your project, if Cloud Composer Service Agent does not have required permissions on the environment's service account, then a Grant required permissions to Cloud Composer service account section appears.

You add Cloud Composer Service Agent account as a new principal on your environment's service account and grant the Cloud Composer v2 API Service Agent Extension role to it.

Confirm that you use the intended service account for your environment, and click Grant.

gcloud

You add Cloud Composer Service Agent account as a new principal on your environment's service account and grant the Cloud Composer v2 API Service Agent Extension role to it.

gcloud iam service-accounts add-iam-policy-binding \
    SERVICE_ACCOUNT \
    --member serviceAccount:service-PROJECT_NUMBER@cloudcomposer-accounts.iam.gserviceaccount.com \
    --role roles/composer.ServiceAgentV2Ext

Replace:

SERVICE_ACCOUNT with the service account for your environment.
PROJECT_NUMBER with the Project number.

Example:

gcloud iam service-accounts add-iam-policy-binding \
    example-account@example-project.iam.gserviceaccount.com \
    --member serviceAccount:service-00000000000@cloudcomposer-accounts.iam.gserviceaccount.com \
    --role roles/composer.ServiceAgentV2Ext

API

To grant the role, you must modify the existing allow policy using the read-modify-write pattern:

Read the existing allow policy for the service account of your environment.
Modify it to include the roles/composer.ServiceAgentV2Ext role for the Cloud Composer Service Agent.
Rewrite the existing allow policy.

For more information, see Controlling access programmatically.

{
  "role": "roles/composer.ServiceAgentV2Ext",
  "members": [
    "serviceAccount:service-PROJECT_NUMBER@cloudcomposer-accounts.iam.gserviceaccount.com"
  ]
}

Replace:

PROJECT_NUMBER with the Project number.

Example:

{
  "role": "roles/composer.ServiceAgentV2Ext",
  "members": [
    "serviceAccount:service-00000000000@cloudcomposer-accounts.iam.gserviceaccount.com"
  ]
}

Terraform

Add a new role binding to your environment's service account allow policy.

You add Cloud Composer Service Agent account as a new principal on your environment's service account and grant the Cloud Composer v2 API Service Agent Extension role to it.

If you are not using Terraform to define your environment's service account allow policy, do not use the following example. Instead, add this binding using other methods.

resource "google_service_account_iam_member" "custom_service_account" {
  provider = google-beta
  service_account_id = "SERVICE_ACCOUNT"
  role = "roles/composer.ServiceAgentV2Ext"
  member = "serviceAccount:service-PROJECT_NUMBER@cloudcomposer-accounts.iam.gserviceaccount.com"
}

Replace:

SERVICE_ACCOUNT with the service account for your environment.
PROJECT_NUMBER with the Project number.

Example:

resource "google_service_account_iam_member" "custom_service_account" {
  provider = google-beta
  service_account_id = "example-account@example-project.iam.gserviceaccount.com"
  role = "roles/composer.ServiceAgentV2Ext"
  member = "serviceAccount:service-00000000000@cloudcomposer-accounts.iam.gserviceaccount.com"
}

Step 4. (Optional) Configure environment scale and performance parameters

To specify the scale and performance configuration for your environment, select the environment size and workloads configuration.

You can change all performance and scale parameters after you create an environment.

Following parameters control the scale and performance:

Environment size. Controls the performance parameters of the managed Cloud Composer infrastructure that includes the Airflow database. Consider selecting a larger environment size if you want to run a large number of DAGs and tasks with higher infrastructure performance. For example, larger environment's size increases the amount of Airflow task log entries that your environment can process with minimal delay.
Workloads configuration. Controls the scale and performance of your environment components that run in a GKE cluster: Airflow schedulers, Airflow web server, and Airflow workers.
- Airflow scheduler. Parses DAG definition files, schedules DAG runs based on the schedule interval, and queues tasks for execution by Airflow workers.
  
  Your environment can run more than one Airflow scheduler at the same time. Use multiple schedulers to distribute load between several scheduler instances for better performance and reliability.
  
  Increasing the number of schedulers does not always improve Airflow performance. For example, having only one scheduler might provide better performance than having two. This might happen when the extra scheduler is not utilized, and thus consumes resources of your environment without contributing to overall performance. The actual scheduler performance depends on the number of Airflow workers, the number of DAGs and tasks that run in your environment, and the configuration of both Airflow and the environment.
  
  We recommend starting with two schedulers and then monitoring the performance of your environment. If you change the number of schedulers, you can always scale your environment back to the original number of schedulers.
  
  For more information about configuring multiple schedulers, see Airflow documentation.
- Airflow triggerer. Asynchronously monitors all deferred tasks in your environment. If you have at least one triggerer instance in your environment (or at least two in highly resilient environments), you can use deferrable operators in your DAGs.
- Airflow web server. Runs the Airflow web interface where you can monitor, manage, and visualize your DAGs.
- Airflow workers. Execute tasks that are scheduled by Airflow schedulers. The minimum and maximum number of workers in your environment changes dynamically depending on the number of tasks in the queue.

Console

You can select a preset for your environment. When you select a preset, the scale and performance parameters for that preset are automatically selected. You also have an option to select a custom preset and specify all scale and performance parameters for your environment.

To select the scale and performance configuration for your environment, on the Create environment page:

To use predefined values, in the Environment resources section, click Small, Medium, or Large.
To specify custom values for the scale and performance parameters:
1. In the Environment resources section, click Custom.
2. In the Scheduler section, set the number of schedulers you want to use, and the resource allocation for their CPU, memory, and storage.
3. In the Triggerer section, use the Number of triggerers field to enter the number of triggerers in your environment. You can set this number to 0 if you don't want to use deferrable operators in your DAGs.
  
  If you set at least one triggerer for your environment, use the the CPU, and Memory fields to configure resource allocation for your triggerers.
4. In the DAG processor section, specify the number of DAG processors in your enviueornment and the amount of CPUs, memory, and storage for each DAG processor.
5. In the Web server section, specify the amount of CPUs, memory, and storage for the web server.
6. In the Worker section, specify:
  - The minimum and maximum number of workers for autoscaling limits in your environment.
  - The CPU, memory, and storage allocation for your workers
7. In the Core infrastructure section, in the Environment size drop-down list, select the environment size.

gcloud

When you create an environment, the following arguments control the scale and performance parameters of your environment.

--environment-size specifies the environment size.
--scheduler-count specifies the number of schedulers.
--scheduler-cpu specifies the number of CPUs for an Airflow scheduler.
--scheduler-memory specifies the amount of memory for an Airflow scheduler.
--scheduler-storage specifies the amount of disk space for an Airflow scheduler.
--triggerer-count specifies the number of Airflow triggerers in your environment. The default value for this flag is 0. You need triggerers if you want to use deferrable operators in your DAGs.
- For standard resilience environments, use a value between 0 and 10.
- For highly resilient environments, use 0 or a value between 2 and 10.
--triggerer-cpu specifies the number of CPUs for an Airflow triggerer, in vCPU units. Allowed values: 0.5, 0.75, 1. The default value is 0.5.
--triggerer-memory specifies the amount of memory for an Airflow triggerer, in GB. The default value is 0.5.

The minimum required memory is equal to the number of CPUs allocated for the triggerers. The maximum allowed value is equal to the number of triggerer CPUs multiplied by 6.5.

For example, if you set the --triggerer-cpu flag to 1, the minimum value for --triggerer-memory is 1 and the maximum value is 6.5.
--web-server-cpu specifies the number of CPUs for the Airflow web server.
--web-server-memory specifies the amount of memory for the Airflow web server.
--web-server-storage specifies the amount of disk space for the Airflow web server.
--worker-cpu specifies the number of CPUs for an Airflow worker.
--worker-memory specifies the amount of memory for an Airflow worker.
--worker-storage specifies the amount of disk space for an Airflow worker.
--min-workers specifies the minimum number of Airflow workers. Your environment's cluster runs at least this number of workers.
--max-workers specifies the maximum number of Airflow workers. Your environment's cluster runs at most this number of workers.

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --environment-size ENVIRONMENT_SIZE \
    --scheduler-count SCHEDULER_COUNT \
    --scheduler-cpu SCHEDULER_CPU \
    --scheduler-memory SCHEDULER_MEMORY \
    --scheduler-storage SCHEDULER_STORAGE \
    --triggerer-count TRIGGERER_COUNT \
    --triggerer-cpu TRIGGERER_CPU \
    --triggerer-memory TRIGGERER_MEMORY \
    --web-server-cpu WEB_SERVER_CPU \
    --web-server-memory WEB_SERVER_MEMORY \
    --web-server-storage WEB_SERVER_STORAGE \
    --worker-cpu WORKER_CPU \
    --worker-memory WORKER_MEMORY \
    --worker-storage WORKER_STORAGE \
    --min-workers WORKERS_MIN \
    --max-workers WORKERS_MAX

Replace:

ENVIRONMENT_SIZE with small, medium, or large.
SCHEDULER_COUNT with the number of schedulers.
SCHEDULER_CPU with the number of CPUs for a scheduler, in vCPU units.
SCHEDULER_MEMORY with the amount of memory for a scheduler.
SCHEDULER_STORAGE with the disk size for a scheduler.
TRIGGERER_COUNT with the number of triggerers.
TRIGGERER_CPU with the number of CPUs for a triggerer, in vCPU units.
TRIGGERER_MEMORY with the amount of memory for a triggerer, in GB.
WEB_SERVER_CPU with the number of CPUs for the web server, in vCPU units.
WEB_SERVER_MEMORY with the amount of memory for the web server.
WEB_SERVER_STORAGE with the amount of memory for the web server.
WORKER_CPU with the number of CPUs for a worker, in vCPU units.
WORKER_MEMORY with the amount of memory for a worker.
WORKER_STORAGE with the disk size for a worker.
WORKERS_MIN with the minimum number of Airflow workers that your environment can run. The number of workers in your environment does not go above this number, even if a lower number of workers can handle the load.
WORKERS_MAX with the maximum number of Airflow workers that your environment can run. The number of workers in your environment does not go above this number, even if a higher number of workers is required to handle the load.

Example:

gcloud composer environments create example-environment \
    --location us-central1 \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --environment-size small \
    --scheduler-count 1 \
    --scheduler-cpu 0.5 \
    --scheduler-memory 2.5GB \
    --scheduler-storage 2GB \
    --triggerer-count 1 \
    --triggerer-cpu 0.5 \
    --triggerer-memory 0.5GB \
    --web-server-cpu 1 \
    --web-server-memory 2.5GB \
    --web-server-storage 2GB \
    --worker-cpu 1 \
    --worker-memory 2GB \
    --worker-storage 2GB \
    --min-workers 2 \
    --max-workers 4

API

When you create an environment, in the Environment > EnvironmentConfig > WorkloadsConfig resource, specify environment scale and performance parameters.

{
  "name": "projects/PROJECT_ID/locations/LOCATION/environments/ENVIRONMENT_NAME",
  "config": {
    "workloadsConfig": {
      "scheduler": {
        "cpu": SCHEDULER_CPU,
        "memoryGb": SCHEDULER_MEMORY,
        "storageGb": SCHEDULER_STORAGE,
        "count": SCHEDULER_COUNT
      },
      "triggerer": {
        "count": TRIGGERER_COUNT,
        "cpu": TRIGGERER_CPU,
        "memoryGb": TRIGGERER_MEMORY
      },
      "webServer": {
        "cpu": WEB_SERVER_CPU,
        "memoryGb": WEB_SERVER_MEMORY,
        "storageGb": WEB_SERVER_STORAGE
      },
      "worker": {
        "cpu": WORKER_CPU,
        "memoryGb": WORKER_MEMORY,
        "storageGb": WORKER_STORAGE,
        "minCount": WORKERS_MIN,
        "maxCount": WORKERS_MAX
      }
    },
    "environmentSize": "ENVIRONMENT_SIZE"
  }
}

Replace:

SCHEDULER_CPU with the number of CPUs for a scheduler, in vCPU units.
SCHEDULER_MEMORY with the amount of memory for a scheduler, in GB.
SCHEDULER_STORAGE with the disk size for a scheduler, in GB.
SCHEDULER_COUNT with the number of schedulers.
TRIGGERER_COUNT with the number of triggerers. The default value is 0. You need triggerers if you want to use deferrable operators in your DAGs.
- For standard resilience environments, use a value between 0 and 10.
- For highly resilient environments, use 0 or a value between 2 and 10.
If you use at least one triggerer, you must also specify the TRIGGERER_CPU, and TRIGGERER_MEMORY parameters:
TRIGGERER_CPU specifies the number of CPUs for a triggerer, in vCPU units. Allowed values: 0.5, 0.75, 1.
TRIGGERER_MEMORY configures the amount of memory for a triggerer. The minimum required memory is equal to the number of CPUs allocated for the triggerers. The maximum allowed value is equal to the number of triggerer CPUs multiplied by 6.5.

For example, if you set the TRIGGERER_CPU to 1, the minimum value for TRIGGERER_MEMORY is 1 and the maximum value is 6.5.
WEB_SERVER_CPU with the number of CPUs for the web server, in vCPU units.
WEB_SERVER_MEMORY with the amount of memory for the web server, in GB.
WEB_SERVER_STORAGE with the disk size for the web server, in GB.
WORKER_CPU with the number of CPUs for a worker, in vCPU units.
WORKER_MEMORY with the amount of memory for a worker, in GB.
WORKER_STORAGE with the disk size for a worker, in GB.
WORKERS_MIN with the minimum number of Airflow workers that your environment can run. The number of workers in your environment does not go above this number, even if a lower number of workers can handle the load.
WORKERS_MAX with the maximum number of Airflow workers that your environment can run. The number of workers in your environment does not go above this number, even if a higher number of workers is required to handle the load.
ENVIRONMENT_SIZE with the environment size, ENVIRONMENT_SIZE_SMALL, ENVIRONMENT_SIZE_MEDIUM, or ENVIRONMENT_SIZE_LARGE.

Example:

// POST https://composer.googleapis.com/v1/{parent=projects/*/locations/*}/environments

{
  "name": "projects/example-project/locations/us-central1/environments/example-environment",
  "config": {
    "workloadsConfig": {
      "scheduler": {
        "cpu": 2.5,
        "memoryGb": 2.5,
        "storageGb": 2,
        "count": 1
      },
      "triggerer": {
        "cpu": 0.5,
        "memoryGb": 0.5,
        "count": 1
      },
      "webServer": {
        "cpu": 1,
        "memoryGb": 2.5,
        "storageGb": 2
      },
      "worker": {
        "cpu": 1,
        "memoryGb": 2,
        "storageGb": 2,
        "minCount": 2,
        "maxCount": 4
      }
    },
    "environmentSize": "ENVIRONMENT_SIZE_SMALL"
  }
}

Terraform

When you create an environment, following arguments control the scale and performance parameters of your environment.

In the config block:
- The environment_size field controls the environment size.
In the workloads_config block:
- The scheduler.cpu field specifies the number of CPUs for an Airflow scheduler.
- The scheduler.memory_gb field specifies the amount of memory for an Airflow scheduler.
- The scheduler.storage_gb field specifies the amount of disk space for a scheduler.
- The scheduler.count field specifies the number of schedulers in your environment.
- The triggerer.cpu field specifies the number of CPUs for an Airflow triggerer.
- The triggerer.memory_gb field specifies the amount of memory for an Airflow triggerer.
- The triggerer.count field specifies the number of triggerers in your environment.
- The web_server.cpu field specifies the number of CPUs for the Airflow web server.
- The web_server.memory_gb field specifies the amount of memory for the Airflow web server.
- The web_server.storage_gb field specifies the amount of disk space for the Airflow web server.
- The worker.cpu field specifies the number of CPUs for an Airflow worker.
- The worker.memory_gb field specifies the amount of memory for an Airflow worker.
- The worker.storage_gb field specifies the amount of disk space for an Airflow worker.
- The worker.min_count field specifies the minimum number of workers in your environment.
- The worker.max_count field specifies the maximum number of workers in your environment.

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "ENVIRONMENT_NAME"
  region = "LOCATION"

  config {

    workloads_config {

      scheduler {
        cpu = SCHEDULER_CPU
        memory_gb = SCHEDULER_MEMORY
        storage_gb = SCHEDULER_STORAGE
        count = SCHEDULER_COUNT
      }
      triggerer {
        count = TRIGGERER_COUNT
        cpu = TRIGGERER_CPU
        memory_gb = TRIGGERER_MEMORY
      }
      web_server {
        cpu = WEB_SERVER_CPU
        memory_gb = WEB_SERVER_MEMORY
        storage_gb = WEB_SERVER_STORAGE
      }
      worker {
        cpu = WORKER_CPU
        memory_gb = WORKER_MEMORY
        storage_gb = WORKER_STORAGE
        min_count = WORKERS_MIN
        max_count = WORKERS_MAX
      }
    }

    environment_size = "ENVIRONMENT_SIZE"

  }
}

Replace:

ENVIRONMENT_NAME with the name of the environment.
LOCATION with the region where the environment is located.
SCHEDULER_CPU with the number of CPUs for a scheduler, in vCPU units.
SCHEDULER_MEMORY with the amount of memory for a scheduler, in GB.
SCHEDULER_STORAGE with the disk size for a scheduler, in GB.
SCHEDULER_COUNT with the number of schedulers.
TRIGGERER_COUNT with the number of triggerers.
TRIGGERER_CPU with the number of CPUs for a triggerer, in vCPU units.
TRIGGERER_MEMORY with the amount of memory for a triggerer, in GB.
WEB_SERVER_CPU with the number of CPUs for the web server, in vCPU units.
WEB_SERVER_MEMORY with the amount of memory for the web server, in GB.
WEB_SERVER_STORAGE with the disk size for the web server, in GB.
WORKER_CPU with the number of CPUs for a worker, in vCPU units.
WORKER_MEMORY with the amount of memory for a worker, in GB.
WORKER_STORAGE with the disk size for a worker, in GB.
WORKERS_MIN with the minimum number of Airflow workers that your environment can run. The number of workers in your environment does not go above this number, even if a lower number of workers can handle the load.
WORKERS_MAX with the maximum number of Airflow workers that your environment can run. The number of workers in your environment does not go above this number, even if a higher number of workers is required to handle the load.
ENVIRONMENT_SIZE with the environment size, ENVIRONMENT_SIZE_SMALL, ENVIRONMENT_SIZE_MEDIUM, or ENVIRONMENT_SIZE_LARGE.

Example:

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "example-environment"
  region = "us-central1"

  config {

    workloads_config {

      scheduler {
        cpu = 2.5
        memory_gb = 2.5
        storage_gb = 2
        count = 1
      }
      triggerer {
        count = 1
        cpu = 0.5
        memory_gb = 0.5
      }
      web_server {
        cpu = 1
        memory_gb = 2.5
        storage_gb = 2
      }
      worker {
        cpu = 1
        memory_gb = 2
        storage_gb = 2
        min_count = 2
        max_count = 4
      }
    }

    environment_size = "ENVIRONMENT_SIZE_SMALL"

  }
}

Step 5. (Optional) Enable high resilience mode

Highly resilient Cloud Composer environments are environments that use built-in redundancy and failover mechanisms that reduce the environment's susceptibility to zonal failures and single point of failure outages.

A highly resilient environment runs across at least two zones of a selected region. Exactly two Airflow schedulers, two web servers, and at least two triggerers (if the number of triggerers isn't set to 0) run in separate zones. The minimum number of workers is set to two, and your environment's cluster distributes worker instances between zones. In case of a zonal outage, affected worker instances are rescheduled in a different zone. The Cloud SQL database of a highly resilient environment is a regional instance that has a primary instance and a standby instance.

Console

On the Create environment page:

In the Resilience mode section, select High resilience.
In the Environment resources section, select scale parameters for a highly resilient environment. Highly resilient environments require exactly two schedulers, zero or between two and ten triggerers, and at least two workers:
1. Click Custom.
2. In the Number of schedulers drop-down list, select 2.
3. In the Number of triggerers drop-down list, select 0, or a value between 2 and 10. Configure the CPU and Memory allocation for your triggerers.
4. In the Minimum number of workers drop-down list, select 2 or more, depending on the required number of workers.
In the Network configuration section:
1. In the Networking type, select Private IP environment.
2. If required, specify other networking parameters.

gcloud

When you create an environment, the --enable-high-resilience argument enables the high resilience mode.

Set the following arguments:

--enable-high-resilience
--enable-private-environment, and other networking parameters for a Private IP environment, if required
--scheduler-count to 2
--triggerer-count to 0 or a value between 2 and 10. If you use triggerers, the --triggerer-cpu and--triggerer-memory` flags are also required for environment creation.

For more information about --triggerer-count, --triggerer-cpu, and --triggerer-memory flags, see Configure environment scale and performance parameters.
--min-workers to 2 or more

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --enable-high-resilience \
    --enable-private-environment \
    --scheduler-count 2 \
    --triggerer-count 2 \
    --triggerer-cpu 0.5 \
    --triggerer-memory 0.5 \
    --min-workers 2

API

When you create an environment, in the Environment > EnvironmentConfig resource, enable the high resilience mode.

{
  "name": "projects/PROJECT_ID/locations/LOCATION/environments/ENVIRONMENT_NAME",
  "config": {
    "resilience_mode": "HIGH_RESILIENCE"
  }
}

Example:


// POST https://composer.googleapis.com/v1/{parent=projects/*/locations/*}/environments

{
  "name": "projects/example-project/locations/us-central1/environments/example-environment",
  "config": {
    "resilience_mode": "HIGH_RESILIENCE"
  }
}

Terraform

When you create an environment, the resilience_mode field in the config block enables the high resilience mode.

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "ENVIRONMENT_NAME"
  region = "LOCATION"

  config {

    resilience_mode = "HIGH_RESILIENCE"

  }
}

Example:

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "example-environment"
  region = "us-central1"

  config {

    resilience_mode = "HIGH_RESILIENCE"

}

Step 6. (Optional) Specify a zone for the environment's database

You can specify a preferred Cloud SQL zone when creating a standard resilience environment.

Console

On the Create environment page:

In the Advanced configuration section, expand the Show advanced configuration item.
In the Airflow database zone list, select a preferred Cloud SQL zone.

gcloud

When you create an environment, the --cloud-sql-preferred-zone argument specifies a preferred Cloud SQL zone.

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --cloud-sql-preferred-zone SQL_ZONE

Replace the following:

SQL_ZONE: preferred Cloud SQL zone. This zone must be located in the region where the environment is located.

Example:

gcloud composer environments create example-environment \
    --location us-central1 \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --cloud-sql-preferred-zone us-central1-a

API

When you create an environment, in the Environment > DatabaseConfig resource, specify the preferred Cloud SQL zone.

{
  "name": "projects/PROJECT_ID/locations/LOCATION/environments/ENVIRONMENT_NAME",
  "config": {
    "databaseConfig": {
      "zone": "SQL_ZONE"
    }
  }
}

Replace the following:

SQL_ZONE: preferred Cloud SQL zone. This zone must be located in the region where the environment is located.

Example:


// POST https://composer.googleapis.com/v1/{parent=projects/*/locations/*}/environments

{
  "name": "projects/example-project/locations/us-central1/environments/example-environment",
  "config": {
    "databaseConfig": {
      "zone": "us-central1-a"
    }
  }
}

Terraform

When you create an environment, the zone field in the database_config block specifies the preferred Cloud SQL zone.

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "ENVIRONMENT_NAME"
  region = "LOCATION"

  config {
    database_config {
      zone = "SQL_ZONE"
    }
  }
}

Replace the following:

SQL_ZONE: preferred Cloud SQL zone. This zone must be located in the region where the environment is located.

Step 7. (Optional) Configure your environment's networking

Networking parameters depend on the type of environment that you want to create:

Public IP environment. Use the default networking parameters.
Private IP environment (with PSC). In this configuration, your environment uses Private Service Connect for connectivity.

Configure your Private IP environment:
1. Configure your project's networking for Private IP environments.
2. Configure Private Service Connect when you create your environment.
3. Specify other parameters for your Private IP environment, as described further in this section.
For a Private IP environment with PSC, you need to know:
- Your VPC network ID
- Your VPC subnetwork ID
- Two secondary IP ranges in your VPC subnetwork:
  - Secondary IP range for pods
  - Secondary IP range for services
- IP ranges for the components of the environment:
  - GKE Control Plane IP range. IP range for the GKE control plane.
  - Cloud Composer connection subnetwork. IP range for the Cloud Composer connection subnetwork.
Private IP environment (VPC peerings). In this configuration, your environment uses VPC peerings for connectivity.

Configure your Private IP environment:
1. Configure your project's networking for Private IP environments.
2. Specify other parameters for your Private IP environment, as described further in this section.
For a Private IP environment with VPC peerings, you need to know:
- Your VPC network ID
- Your VPC subnetwork ID
- Two secondary IP ranges in your VPC subnetwork:
  - Secondary IP range for pods
  - Secondary IP range for services
- IP ranges for the components of the environment:
  - IP range for the GKE control plane.
  - IP range for VPC peering to export from the internal Cloud Composer network to the selected network. Cloud Composer infrastructure components use IP addresses from this range.
  - IP range for the Cloud SQL instance.
For a Shared VPC environment, you must do additional networking setup for the host project, then create a Public or a Private IP environment in a service project. Follow the instructions on the Configuring Shared VPC page.

Important: If you create an environment in the service project using Google Cloud console, then your account must have the compute.subnetworks.use permission in the host project. Otherwise, the list of available subnetworks does not contain subnetworks from the host project. If you create an environment using gcloud, API, or Terraform, then your account does not need this additional permission.

For a Shared VPC environment, you need to know:
- Your host project VPC network ID
- Your host project VPC subnetwork ID
- Two secondary IP ranges in your host project VPC subnetwork:
  - Secondary IP range for pods
  - Secondary IP range for services
When creating a Public IP Shared VPC environment, you still need to specify your host project VPC network, subnetwork, and secondary IP ranges for pods and services.
To create a VPC SC environment, you must create a service perimeter and then create Private IP environments inside this perimeter. Follow the instructions outlined in Configuring VPC Service Controls.

Additional networking options for environments are:

Privately used public IP addresses. If you want to use more IP addresses, your environment can privately use certain public IP address ranges as internal, subnet IP address ranges for pods and services.
Authorized networks. If you want to access the control plane of your Private IP environment using HTTPS, you can use authorized networks to specify CIDR ranges that can do so.
IP Masquerade agent. By using environments with the IP Masquerade agent, you can use many-to-one IP address translations in your environment's networking configurations. For more information about creating environments with IP Masquerade agent, see Enable the IP Masquerade agent.

Console

To create a Private IP environment:

Make sure that your networking is configured for the type of environment that you want to create.
In the Network configuration section, expand the Show network configuration item.
In the Network drop-down list, select your VPC network ID.
In the Subnetwork drop-down list, select your VPC subnetwork ID.
In the Secondary IP range for pods section, select or specify the secondary IP range for pods. You can use an existing secondary range in your VPC network, or choose to use an auto-created range.
In the Secondary IP range for services section, select or specify the secondary IP range for services. You can use an existing secondary range in your VPC network, or choose to use an auto-created range.
In the Networking type section, select the Private IP environment option to create a Private IP environment.
In the Composer connectivity section, select the networking type for your environment and specify IP ranges for environment components:

For an environment that uses Private Service Connect:
1. Select Private Service Connect for an environment that uses Private Service Connect.
2. In the Composer connection subnetwork section specify an IP range for the Cloud Composer connection subnetwork. The address for the PSC endpoint is selected from this range. You can specify a custom range, or choose to use the default one.
For an environment that uses VPC peerings:
1. Select VPC peerings for an environment that uses VPC peerings.
  
  Note: Environments with VPC peerings have more limitations than environments with PSC.
2. In the IP range for Composer tenant network section specify an IP range for the Cloud Composer tenant network. This network hosts the SQL proxy component of your environment. You can specify a custom range, or choose to use the default one.
3. In the IP range for Cloud SQL network section, specify an IP range for the Cloud SQL instance. You can specify a custom range, or choose to use the default one.
  
  Caution: The 172.17.0.0/16 range is reserved in Cloud SQL.
In the IP range for GKE control plane network section, specify an IP range for the GKE control plane:
- To use the default IP range for the region where your environment is located, select Default IP range.
- To specify a custom IP range, select Custom IP range and enter a range in the CIDR notation in the GKE cluster master private IP field.
Select the level access for the GKE control plane. The control plane has two endpoints. One endpoint is private, for use by cluster nodes and VMs. Another endpoint is public. You can specify the level of access for the public endpoint:
- To enable access to the public endpoint from authorized networks, select the Access cluster control plane endpoint using its external IP address checkbox.
  
  Using this option sets the level of access for the control plane to "Public endpoint access enabled, authorized networks enabled". This provides restricted access to the control plane from authorized networks. By default, no source IP addresses are specified. You can add authorized networks to the cluster.
- To disable access to the public endpoint from authorized networks, clear the Access cluster control plane endpoint using its external IP address checkbox.
  
  Using this option sets the level of access for the control plane to "Public endpoint access disabled". This prevents all internet access to the control plane.

gcloud

Make sure that your networking is configured for the type of environment that you want to create.

When you create an environment, the following arguments control the networking parameters. If you omit a parameter, the default value is used.

--enable-private-environment enables a Private IP environment.
--network specifies your VPC network ID.
--subnetwork specifies your VPC subnetwork ID.
--cluster-secondary-range-name or --cluster-ipv4-cidr configures the the secondary range for pods.
--services-secondary-range-name or--services-ipv4-cidr to configure the secondary range for services.
--master-ipv4-cidr specifies a range for the GKE control plane.

(Environments with PSC) --connection-subnetwork specifies a range for the Cloud Composer connection subnetwork, which hosts the PSC endpoint.
(Environments with VPC peerings) --composer-network-ipv4-cidr specifies a range for the Cloud Composer tenant network. This network hosts the SQL proxy component of your environment.
(Environments with VPC peerings) --cloud-sql-ipv4-cidr specifies a range for the Cloud SQL instance.

--enable-private-endpoint controls the level access for the GKE control plane. The control plane has two endpoints. One endpoint is private, for use by cluster nodes and VMs. Another endpoint is public. You can specify the level of access for the public endpoint:
- To enable access to the public endpoint from authorized networks, omit the --enable-private-endpoint argument.
  
  Using this option sets the level of access for the control plane to "Public endpoint access enabled, authorized networks enabled". This provides restricted access to the control plane from authorized networks. By default, no source IP addresses are specified. You can add authorized networks to the cluster.
- To disable access to the public endpoint from authorized networks, specify the --enable-private-endpoint argument.
  
  Using this option sets the level of access for the control plane to "Public endpoint access disabled". This prevents all internet access to the control plane.
--enable-master-authorized-networks and --master-authorized-networks arguments configure authorized networks for your environment.
--enable-privately-used-public-ips configures privately used public IP addresses for your environment.
--enable-ip-masq-agent enables the IP Masquerade agent.

Example (Private IP environment )

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --enable-private-environment \
    --network NETWORK_ID \
    --subnetwork SUBNETWORK_ID \
    --cluster-ipv4-cidr PODS_RANGE \
    --services-ipv4-cidr SERVICES_RANGE \
    --master-ipv4-cidr CONTROL_PLANE_RANGE \
    --connection-subnetwork COMPOSER_PSC_RANGE \

Replace:

NETWORK_ID with your VPC network ID.
SUBNETWORK_ID with your VPC subnetwork ID.
PODS_RANGE with the secondary range for pods.
SERVICES_RANGE with the secondary range for services.
CONTROL_PLANE_RANGE with the secondary range for the GKE control plane.
COMPOSER_PSC_RANGE with the range for the Cloud Composer connection subnetwork.

Step 8. (Optional) Add network tags

Network tags are applied to all node VMs in your environment's cluster. Tags are used to identify valid sources or targets for network firewalls. Each tag within the list must comply with RFC 1035.

For example, you might want to add network tags if you plan to restrict traffic for a Private IP environment with firewall rules.

Console

On the Create environment page:

Locate the Network configuration section.
In the Network tags field, enter network tags for your environment.

gcloud

When you create an environment, following arguments control network tags:

--tags specifies a comma-separated list of network tags applied to all node VMs.

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --tags TAGS

Replace:

TAGS with a comma-separated list of network tags.

Example:

gcloud composer environments create example-environment \
    --location us-central1 \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --tags group1,production

API

When you create an environment, in the Environment > EnvironmentConfig resource, specify network tags for your environment.

{
  "name": "projects/PROJECT_ID/locations/LOCATION/environments/ENVIRONMENT_NAME",
  "config": {
    "nodeConfig": {
      "tags": [
        "TAG"
      ]
    }
  }
}

Replace:

TAG with a network tag.

Example:

// POST https://composer.googleapis.com/v1/{parent=projects/*/locations/*}/environments

{
  "name": "projects/example-project/locations/us-central1/environments/example-environment",
  "config": {
    "nodeConfig": {
      "tags": [
        "group1",
        "production"
      ]
    }
  }
}

Terraform

When you create an environment, following fields define network tags for your environment:

tags field in the node_config block specifies a comma-separated list of network tags applied to all node VMs.

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "ENVIRONMENT_NAME"
  region = "LOCATION"

  config {

    node_config {
      tags = ["TAGS"]
    }
  }
}

Replace:

TAGS with a comma-separated list of network tags.

Example:

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "example-environment"
  region = "us-central1"

  config {
    node_config {
      tags = ["group1","production"]
    }
  }
}

Step 9. (Optional) Configure web server network access

The Airflow web server access parameters do not depend on the type of your environment. Instead, you can configure web server access separately. For example, a Private IP environment can still have the Airflow UI accessible from the internet.

You cannot configure the allowed IP ranges using private IP addresses.

Console

On the Create environment page:

In the Network configuration section, expand the Show network configuration item.
In the Web server network access control section:
- To provide access to the Airflow web server from all IP addresses, select Allow access from all IP addresses.
- To restrict access only to specific IP ranges, select Allow access only from specific IP addresses. In the IP range field, specify an IP range in the CIDR notation. In the Description field, specify an optional description for this range. If you want to specify more than one range, click Add IP range.
- To forbid access for all IP addresses, select Allow access only from specific IP addresses and click Delete item next to the empty range entry.

gcloud

When you create an environment, following arguments control web server access level:

--web-server-allow-all provides access to Airflow from all IP addresses. This is the default option.
--web-server-allow-ip restricts access only to specific source IP ranges. To specify several IP ranges, use this argument multiple times.
--web-server-deny-all forbids access for all IP addresses.

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --web-server-allow-ip ip_range=WS_IP_RANGE,description=WS_RANGE_DESCRIPTION

Replace:

WS_IP_RANGE with the IP range, in the CIDR notation, that can access Airflow UI.
WS_RANGE_DESCRIPTION with the description of the IP range.

Example:

gcloud composer environments create example-environment \
    --location us-central1 \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --web-server-allow-ip ip_range=192.0.2.0/24,description="office net 1" \
    --web-server-allow-ip ip_range=192.0.4.0/24,description="office net 3"

API

When you create an environment, in the Environment > EnvironmentConfig resource, specify web server access parameters.

To provide access to the Airflow web server from all IP addresses, omit webServerNetworkAccessControl.
To restrict access only to specific IP ranges, specify one or more ranges in allowedIpRanges.
To forbid access for all IP addresses, add allowedIpRanges and make it an empty list. Do not specify IP ranges in it.

{
  "name": "projects/PROJECT_ID/locations/LOCATION/environments/ENVIRONMENT_NAME",
  "config": {
    "webServerNetworkAccessControl": {
      "allowedIpRanges": [
        {
          "value": "WS_IP_RANGE",
          "description": "WS_RANGE_DESCRIPTION"
        }
      ]
    }
  }
}

Replace:

WS_IP_RANGE with the IP range, in the CIDR notation, that can access Airflow UI.
WS_RANGE_DESCRIPTION with the description of the IP range.

Example:


// POST https://composer.googleapis.com/v1/{parent=projects/*/locations/*}/environments

{
  "name": "projects/example-project/locations/us-central1/environments/example-environment",
  "config": {
    "webServerNetworkAccessControl": {
      "allowedIpRanges": [
        {
          "value": "192.0.2.0/24",
          "description": "office net 1"
        },
        {
          "value": "192.0.4.0/24",
          "description": "office net 3"
        }
      ]
    }
  }
}

Terraform

When you create an environment, the allowed_ip_range block in the web_server_network_access_control block contains IP ranges that can access web server.

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "ENVIRONMENT_NAME"
  region = "LOCATION"

  config {

    web_server_network_access_control {

      allowed_ip_range {
        value = "WS_IP_RANGE"
        description = "WS_RANGE_DESCRIPTION"
      }

    }

  }
}

Replace:

WS_IP_RANGE with the IP range, in the CIDR notation, that can access Airflow UI.
WS_RANGE_DESCRIPTION with the description of the IP range.

Example:

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "example-environment"
  region = "us-central1"

  config {

    web_server_network_access_control {
      allowed_ip_range {
        value = "192.0.2.0/24"
        description = "office net 1"
      },
      allowed_ip_range {
        value = "192.0.4.0/24"
        description = "office net 3"
      }

    }
}

Step 10. (Optional) Specify Airflow configuration overrides and environment variables

You can set up Airflow configuration overrides and environment variables when you create an environment. As an alternative, you can do it later, after your environment is created.

Some Airflow configuration options are blocked and you cannot override them.

For the list of available Airflow configuration options, see Configuration reference for Airflow 2 and Airflow 1.10.*

To specify Airflow configuration overrides and environment variables:

Console

On the Create environment page:

In the Environment variables section, click Add environment variable.
Enter the Name and Value for the environment variable.
In the Airflow configuration overrides section, click Add Airflow configuration override.
Enter the Section, Key, and Value for the configuration option override.

For example:

Section Key Value

webserver dag_orientation TB

Section	Key	Value
`webserver`	`dag_orientation`	`TB`

gcloud

When you create an environment, following arguments control environment variables and Airflow configuration overrides:

--env-variables specifies a comma-separated list of environment variables.

Variable names may contain upper and lowercase letters, digits, and underscores, but they may not begin with a digit.
--airflow-configs specifies a comma-separated list of keys and values for Airflow configuration overrides.

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --env-variables ENV_VARS \
    --airflow-configs CONFIG_OVERRIDES

Replace:

ENV_VARS with a list of comma-separated NAME=VALUE pairs for environment variables.
CONFIG_OVERRIDES with a list of comma-separated SECTION-KEY=VALUE pairs for configuration overrides. Separate the name of the configuration section with a - symbol, followed by the key name. For example: core-dags_are_paused_at_creation.

Example:

gcloud composer environments create example-environment \
    --location us-central1 \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --env-variables SENDGRID_MAIL_FROM=user@example.com,SENDGRID_API_KEY=example-key \
    --airflow-configs core-dags_are_paused_at_creation=True,webserver-dag_orientation=TB

API

When you create an environment, in the Environment > EnvironmentConfig resource, specify environment variables and Airflow configuration overrides.

{
  "name": "projects/PROJECT_ID/locations/LOCATION/environments/ENVIRONMENT_NAME",
  "config": {
    "softwareConfig": {
      "airflowConfigOverrides": {
        "SECTION-KEY": "OVERRIDE_VALUE"
      },
      "envVariables": {
        "VAR_NAME": "VAR_VALUE",
      }
    }
  }
}

Replace:

SECTION with the section in the configuration file where the Airflow configuration option is located.
KEY with the name of the Airflow configuration option.
OVERRIDE_VALUE with a value of the Airflow configuration option.
VAR_NAME with the name of the environment variable.
VAR_VALUE with the value of the environment variable.

Example:

// POST https://composer.googleapis.com/v1/{parent=projects/*/locations/*}/environments

{
  "name": "projects/example-project/locations/us-central1/environments/example-environment",
  "config": {
    "softwareConfig": {
      "airflowConfigOverrides": {
        "core-dags_are_paused_at_creation": "True",
        "webserver-dag_orientation": "TB"
      },
      "envVariables": {
        "SENDGRID_MAIL_FROM": "user@example.com",
        "SENDGRID_API_KEY": "example-key"
      }
    }
  }
}

Terraform

When you create an environment, following blocks control environment variables and Airflow configuration overrides:

env_variables block in the software_config block specifies environment variables.

Variable names may contain upper and lowercase letters, digits, and underscores, but they may not begin with a digit.
airflow_config_overrides block in the software_config block specifies Airflow configuration overrides.

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "ENVIRONMENT_NAME"
  region = "LOCATION"

  config {

    software_config {

      airflow_config_overrides = {
        SECTION-KEY = "OVERRIDE_VALUE"
      }

      env_variables = {
        VAR_NAME = "VAR_VALUE"
      }
    }
  }
}

Replace:

SECTION with the section in the configuration file where the Airflow configuration option is located.
KEY with the name of the Airflow configuration option.
OVERRIDE_VALUE with a value of the Airflow configuration option.
VAR_NAME with the name of the environment variable.
VAR_VALUE with the value of the environment variable.

Example:

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "example-environment"
  region = "us-central1"

  config {

    software_config {

      airflow_config_overrides = {
        core-dags_are_paused_at_creation = "True"
        webserver-dag_orientation = "TB"
      }

      env_variables = {
        SENDGRID_MAIL_FROM = "user@example.com"
        SENDGRID_API_KEY = "example-key"
      }
    }
  }
}

Step 11. (Optional) Specify maintenance windows

Default maintenance windows in Cloud Composer 2 are from 00:00:00 to 04:00:00 (GMT) on Friday, Saturday, and Sunday every week.

To specify custom maintenance windows for your environment:

Console

On the Create environment page

Locate the Maintenance windows section.
In the Timezone drop-down list, choose a time zone for maintenance windows.
Set Start time, Days, and Length, so that combined time for the specified schedule is at least 12 hours in a 7-day rolling window. For example, a period of 4 hours every Monday, Wednesday, and Friday provides the required amount of time.

gcloud

The following arguments define maintenance windows parameters:

--maintenance-window-start sets the start time of a maintenance window.
--maintenance-window-end sets the end time of a maintenance window.
--maintenance-window-recurrence sets the maintenance window recurrence.

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --maintenance-window-start 'DATETIME_START' \
    --maintenance-window-end 'DATETIME_END' \
    --maintenance-window-recurrence 'MAINTENANCE_RECURRENCE'

Replace:

ENVIRONMENT_NAME with the name of the environment.
DATETIME_START with the start date and time in the date/time input format. Only the specified time of the day is used, the specified date is ignored.
DATETIME_END with the end date and time in the date/time input format. Only the specified time of the day is used, the specified date is ignored. The specified date and time must be after the start date.
MAINTENANCE_RECURRENCE with an RFC 5545 RRULE for maintenance windows recurrence. Cloud Composer supports two formats:
The FREQ=DAILY format specifies a daily recurrence.
The FREQ=WEEKLY;BYDAY=SU,MO,TU,WE,TH,FR,SA format specifies a recurrence on selected days of the week.

The following example specifies a 6-hour maintenance window between 01:00 and 07:00 (UTC) on Wednesdays, Saturdays, and Sundays. The 1 January, 2023 date is ignored.

gcloud composer environments create example-environment \
  --location us-central1 \
  --image-version composer-2.9.11-airflow-2.9.3 \
  --maintenance-window-start '2023-01-01T01:00:00Z' \
  --maintenance-window-end '2023-01-01T07:00:00Z' \
  --maintenance-window-recurrence 'FREQ=WEEKLY;BYDAY=SU,WE,SA'

API

When you create an environment, in the Environment > EnvironmentConfig resource, specify maintenance windows parameters:

{
  "name": "projects/PROJECT_ID/locations/LOCATION/environments/ENVIRONMENT_NAME",
  "config": {
    "maintenanceWindow": {
        "startTime": "DATETIME_START",
        "endTime": "DATETIME_END",
        "recurrence": "MAINTENANCE_RECURRENCE"
    }
  }
}

Replace:

DATETIME_START with the start date and time in the date/time input format. Only the specified time of the day is used, the specified date is ignored.
DATETIME_END with the end date and time in the date/time input format. Only the specified time of the day is used, the specified date is ignored. The specified date and time must be after the start date.
MAINTENANCE_RECURRENCE with an RFC 5545 RRULE for maintenance windows recurrence. Cloud Composer supports two formats:
The FREQ=DAILY format specifies a daily recurrence.
The FREQ=WEEKLY;BYDAY=SU,MO,TU,WE,TH,FR,SA format specifies a recurrence on selected days of the week.

The following example specifies a 6-hour maintenance window between 01:00 and 07:00 (UTC) on Wednesdays, Saturdays, and Sundays. The 1 January, 2023 date is ignored.

Example:

// POST https://composer.googleapis.com/v1/{parent=projects/*/locations/*}/environments

{
  "name": "projects/example-project/locations/us-central1/environments/example-environment",
  "config": {
    "maintenanceWindow": {
        "startTime": "2023-01-01T01:00:00Z",
        "endTime": "2023-01-01T07:00:00Z",
        "recurrence": "FREQ=WEEKLY;BYDAY=SU,WE,SA"
    }
  }
}

Terraform

The maintenance_window block specifies the maintenance windows for your environment:

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "ENVIRONMENT_NAME"
  region = "LOCATION"

  config {
    maintenance_window {
      start_time = "DATETIME_START"
      end_time = "DATETIME_END"
      recurrence = "MAINTENANCE_RECURRENCE"
    }
  }
}

Replace:

DATETIME_START with the start date and time in the date/time input format. Only the specified time of the day is used, the specified date is ignored.
DATETIME_END with the end date and time in the date/time input format. Only the specified time of the day is used, the specified date is ignored. The specified date and time must be after the start date.
MAINTENANCE_RECURRENCE with an RFC 5545 RRULE for maintenance windows recurrence. Cloud Composer supports two formats:
- The FREQ=DAILY format specifies a daily recurrence.
- The FREQ=WEEKLY;BYDAY=SU,MO,TU,WE,TH,FR,SA format specifies a recurrence on selected days of the week.

The following example specifies a 6-hour maintenance window between 01:00 and 07:00 (UTC) on Wednesdays, Saturdays, and Sundays. The 1 January, 2023 date is ignored.

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "example-environment"
  region = "us-central1"

  config {
    maintenance_window {
      start_time = "2023-01-01T01:00:00Z"
      end_time = "2023-01-01T07:00:00Z"
      recurrence = "FREQ=WEEKLY;BYDAY=SU,WE,SA"
    }
  }
}

Step 12. (Optional) Data lineage integration

Data lineage is a Dataplex feature that lets you track data movement.

Data lineage integration is available in Cloud Composer 2 versions 2.1.2 and later with Airflow versions 2.2.5 and later.

Data lineage integration is automatically enabled in a new Cloud Composer environment if the following conditions are met:

Data Lineage API is enabled in your project. For more information, see Enabling Data Lineage API in Dataplex documentation.

Caution: After you enable Data Lineage API, Dataplex automatically starts ingesting data for BigQuery, Cloud Data Fusion, Dataproc. This happens even if you do not create a Cloud Composer environment with data lineage integration.
A custom Lineage Backend is not configured in Airflow.

You can disable data lineage integration when you create an environment. For example, if you want to override the automatic behavior or choose to enable data lineage later, after the environment is created.

Console

To disable Data lineage integration, on the Create environment page:

In the Advanced configuration section, expand the Show advanced configuration item.
In the Dataplex data lineage integration section, select Disable integration with Dataplex data lineage.

gcloud

When you create an environment, the --disable-cloud-data-lineage-integration argument disables the data lineage integration.

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --disable-cloud-data-lineage-integration

Replace:

ENVIRONMENT_NAME with the name of the environment.
LOCATION with the region where the environment is located.

Example:

gcloud composer environments create example-environment \
    --location us-central1 \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --disable-cloud-data-lineage-integration

Step 13. (Optional) Configure data encryption (CMEK)

By default, data in your environment is encrypted with a key provided by Google.

To use customer-managed encryption keys (CMEK) to encrypt data in your environment, follow the instructions outlined in Using customer-managed encryption keys.

Step 14. (Optional) Use a custom environment's bucket

When you create an environment, Cloud Composer creates a bucket for your environment automatically.

As an alternative, you can specify a custom Cloud Storage bucket from your project. Your environment uses this bucket in the same way as the automatically created bucket.

To use a custom environment bucket, follow the instructions outlined in Use a custom environment's bucket.

Step 15. (Optional) Specify environment labels

You can assign labels to your environments to break down billing costs based on these labels.

Console

On the Create environment page, in the Labels section:

Click Add label.
In Key and Value fields, specify key and value pairs for the environment labels.

gcloud

When you create an environment, the --labels argument specifies a comma-separated list of keys and values with environment labels.

gcloud composer environments create ENVIRONMENT_NAME \
    --location LOCATION \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --labels LABELS

Replace:

LABELS with a list of comma-separated KEY=VALUE pairs for environment labels.

Example:

gcloud composer environments create example-environment \
    --location us-central1 \
    --image-version composer-2.9.11-airflow-2.9.3 \
    --labels owner=engineering-team,env=production

API

When you create an environment, in the Environment resource, specify labels for your environment.

{
  "name": "projects/PROJECT_ID/locations/LOCATION/environments/ENVIRONMENT_NAME",
  "labels": {
    "LABEL_KEY": "LABEL_VALUE"
  }
}

Replace:

LABEL_KEY with a key of the environment label.
LABEL_VALUE with a value of the environment label.

Example:


// POST https://composer.googleapis.com/v1/{parent=projects/*/locations/*}/environments

{
  "name": "projects/example-project/locations/us-central1/environments/example-environment",
  "labels": {
    "owner": "engineering-team",
    "env": "production"
  }
}

Terraform

When you create an environment, specify labels in the labels block ( outside of the config block).

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "ENVIRONMENT_NAME"
  region = "LOCATION"

  labels = {
    LABEL_KEY = "LABEL_VALUE"
  }

}

Replace:

LABEL_KEY with a key of the environment label.
LABEL_VALUE with a value of the environment label.

Example:

resource "google_composer_environment" "example" {
  provider = google-beta
  name = "example-environment"
  region = "us-central1"

  labels = {
    owner = "engineering-team"
    env = "production"
  }

}

Create Cloud Composer environments

Before you begin

Step 1. Basic setup

Console

gcloud

API

Terraform

Step 2. (Optional) Select a service account for your environment

Console

gcloud

API

Terraform

Step 3. Grant required permissions to Cloud Composer service account

Console

gcloud

API

Terraform

Step 4. (Optional) Configure environment scale and performance parameters

Console

gcloud

API

Terraform

Step 5. (Optional) Enable high resilience mode

Console

gcloud

API

Terraform

Step 6. (Optional) Specify a zone for the environment's database

Console

gcloud

API

Terraform

Step 7. (Optional) Configure your environment's networking

Console

gcloud

Step 8. (Optional) Add network tags

Console

gcloud

API

Terraform

Step 9. (Optional) Configure web server network access

Console

gcloud

API

Terraform

Step 10. (Optional) Specify Airflow configuration overrides and environment variables

Console

gcloud

API

Terraform

Step 11. (Optional) Specify maintenance windows

Console

gcloud

API

Terraform

Step 12. (Optional) Data lineage integration

Console

gcloud

Step 13. (Optional) Configure data encryption (CMEK)

Step 14. (Optional) Use a custom environment's bucket

Step 15. (Optional) Specify environment labels

Console

gcloud

API

Terraform

What's next