Upgrade environments

Managed Airflow (Gen 3) | Managed Airflow (Gen 2) | Managed Airflow (Legacy Gen 1)

This page describes how to upgrade your environment to a new Managed Airflow or Airflow version.

About upgrade operations

You can upgrade the Managed Airflow version, the Airflow version, or both at the same time.

When you change the version of Airflow or Managed Airflow used by your environment:

  • Managed Airflow re-deploys your environment's components using new versions of Managed Airflow images.

  • Managed Airflow applies Airflow configuration changes, such as custom PyPI packages or Airflow configuration option overrides, if your environment had them before the upgrade.

  • Managed Airflow updates the Airflow airflow_db connection to point to the new Cloud SQL database.

Upgrading doesn't change how you connect to the resources in your environment, such as the URL of your environment's bucket, or Airflow web server.

Limitations of upgrade operations

Upgrade operations have the following limitations:

  • Managed Airflow releases are gradually rolled out to all regions supported by Managed Airflow over several days. The latest version from a release in progress might not be available in your region yet.

  • You can't downgrade to an earlier Managed Airflow or Airflow version.

  • In Google Cloud console, you can upgrade only to the latest supported versions of Managed Airflow (Gen 2).

  • In Google Cloud CLI, API, or Terraform, you can upgrade an environment to the latest supported versions, three previous versions of Managed Airflow (Gen 2), the last patch version of the previous minor version, and to versions with an extended upgrade timeline. It's not possible to upgrade to other versions of Managed Airflow (Gen 2), even if they are still supported and can be used to create a new environment.

    Example of upgrade choices for Google Cloud CLI, API, and Terraform:

    • Latest versions: composer-2.11.5-airflow-*
    • Three previous Managed Airflow (Gen 2) versions: composer-2.11.4-airflow-*, composer-2.11.3-airflow-*, composer-2.11.2-airflow-*.
    • Last patch of the previous Managed Airflow (Gen 2) minor version: composer-2.10.2-airflow-*
    • Managed Airflow (Gen 2) versions with an extended upgrade timeline: composer-2.10.2-airflow-*, composer-2.10.1-ariflow-*, composer-2.9.7-airflow-*, and other versions with an extended upgrade timeline.

  • The version that you upgrade to must support your environment's current Python version.

  • The Compute Engine API quota limits can prevent upgrades. In Google Cloud console, navigate to the IAM and Admin > Quotas and System Limits page and check if the Compute Engine API quota for CPU is not exceeded. If the quota threshold is being approached, request Quota extension before proceeding with upgrade operation.

  • You can't upgrade your environment if the Airflow database contains more than 16 GB of data. During an upgrade, a warning is shown if the Airflow database size is more than 16 GB. In this case, perform the database maintenance to reduce the database size.

  • If you use the XCom mechanism to transfer files, make sure that you use it according to Airflow's guidelines. Transferring big files or a large number of files using XCom impacts Airflow database's performance and can lead to failures when loading snapshots or upgrading your environment. Consider using alternatives such as Cloud Storage to transfer large volumes of data.

Before you begin

  • We recommend to create a new snapshot of the environment to be able to re-create the environment in case this is needed.

  • Your account must have a role that can trigger environment upgrade operations. In addition, the service account of the environment must have a role that has enough permissions to perform upgrade operations. For more information, see Access control.

  • You might notice Compute Engine metadata entries for the GKE clusters that belong to environments in your project. During GKE cluster creation and upgrades, Google Kubernetes Engine automatically adds project-level metadata entries (google_compute_project_metadata) to track secondary IP address range usage. Don't modify or remove these entries. Google Kubernetes Engine manages them automatically.

Check that your environment is up to date

Managed Airflow displays warnings when your environment's image approaches its end of full support date. You can use these warnings to always keep your environment supported.

A deprecation message is displayed on the Environment details page
Figure 1. A deprecation message is displayed on the Environment details page

Managed Airflow keeps track of the Managed Airflow image version that your environment is based on. When it approaches the end of support date, you can see a warning in the list of environments and on the Environment details page.

To check if your environment is up to date:

Console

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

    Go to Environments

  2. In the list of environments, click the name of your environment. The Environment details page opens.

  3. Go to the Environment configuration tab.

  4. In the Image version field, one of the following messages is displayed:

    • Newest available version. Your environment image is fully supported.

    • New version available. Your environment image is fully supported and you can upgrade it to a later version.

    • Support for this image version ends in... Your environment image approaches the end of the full support period.

    • This version is not supported as of... Your environment is past the full support period.

gcloud

This functionality is not available through Google Cloud CLI. You can View suggested upgrades instead, which shows new versions that are available.

API

This functionality is not available through API. You can View suggested upgrades instead, which shows new versions that are available.

View suggested upgrades

Managed Airflow provides a list of Managed Airflow versions that you can upgrade your environment to.

To view Managed Airflow versions suggested for an upgrade:

Console

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

    Go to Environments

  2. In the list of environments, click the name of your environment. The Environment details page opens.

  3. Go to the Environment configuration tab and click Upgrade Image Version.

  4. For the list of suggested versions, click the Managed Airflow Image version drop-down menu.

gcloud 

gcloud composer environments list-upgrades \
  ENVIRONMENT_NAME \
  --location LOCATION

Replace:

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

Example:

gcloud composer environments list-upgrades example-environment \
  --location us-central1

API

You can view available versions for a location. To do so, construct an imageVersions.list API request.

For example:

// GET https://composer.googleapis.com/v1/projects/example-project/
// locations/us-central1/imageVersions

Check for PyPI package conflicts

You can check if PyPI packages installed in your environment have any conflicts with preinstalled packages in the new Managed Airflow image.

A successful check means that there are no conflicts in PyPI package dependencies between the current and the specified version. However, an upgrade operation might still not be successful because of other reasons.

Console

To run an upgrade check for your environment:

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

    Go to Environments

  2. In the list of environments, click the name of your environment. The Environment details page opens.

  3. Go to the Environment configuration tab, locate the Image version entry, and click Upgrade.

  4. In the Environment version upgrade dialog, in the New version drop-down list, select a Managed Airflow version that you want to upgrade to.

  5. In the PyPI packages compatibility section, click Check for conflicts.

  6. Wait until the check is complete. If there are PyPI package dependency conflicts, then the displayed error messages contain details about conflicting packages and package versions.

gcloud

To run an upgrade check for your environment, run the environments check-upgrade command with the Managed Airflow image version. that you want to upgrade to.

gcloud composer environments check-upgrade \
  ENVIRONMENT_NAME \
  --location LOCATION \
  --image-version VERSION

Replace:

  • ENVIRONMENT_NAME with the name of the environment.
  • LOCATION with the region where the environment is located.
  • VERSION with the new Managed Airflow image version that you want to upgrade to, in the composer-a.b.c-airflow-x.y.z format. You can use all version aliases.

Example:

gcloud composer environments check-upgrade example-environment \
  --location us-central1 \
  --image-version composer-2.16.10-airflow-2.10.5

Example output:

Waiting for [projects/example-project/locations/us-central1/environments/
example-environment] to be checked for PyPI package conflicts when upgrading
to composer-2.16.10-airflow-2.10.5. Operation [projects/example-project/locations/
us-central1/operations/04d0e8b2-...]...done.
...

Response:
'@type': type.googleapis.com/
google.cloud.orchestration.airflow.service.v1.CheckUpgradeResponse
buildLogUri: ...
containsPypiModulesConflict: CONFLICT
pypiConflictBuildLogExtract: |-
The Cloud Build image build failed: Build failed; check build logs for
details. Full log can be found at ...
Error details: tensorboard 2.2.2 has requirement
setuptools>=41.0.0, but you have setuptools 40.3.0.

As an alternative, you can run an upgrade check asynchronously. Use the --async argument to make an asynchronous call, then check the result with the gcloud composer operations describe command.

API

Construct an environments.checkUpgrade API request.

Specify the image version in the imageVersion field:

{
  "imageVersion": "VERSION"
}

Replace VERSION with the new version that you want to upgrade to, in the composer-a.b.c-airflow-x.y.z format.

Check for conflicts with Managed Airflow (Gen 3)

While it isn't possible to migrate your environment from Managed Airflow (Gen 2) to Managed Airflow (Gen 3) in-place, you can check for compatibility problems using an upgrade check. We recommend to do it before migrating to Managed Airflow (Gen 3).

To check for compatibility with Managed Airflow (Gen 3), make a regular upgrade check of your Managed Airflow (Gen 2) environment and specify a Managed Airflow (Gen 3) version as a target.

Each reported conflict can have the following types:

  • BLOCKING conflicts report incompatible environment configuration elements that aren't supported in Managed Airflow (Gen 3). We recommend to address these conflicts before starting a migration to Managed Airflow (Gen 3).

  • NON_BLOCKING conflicts report environment configuration elements that are different in Managed Airflow (Gen 3) and might cause unexpected behavior. For example, changes of the default values such as default maintenance windows are reported as non-blocking conflicts. We recommend to check if these conflicts can cause problems before starting the migration process.

  • PyPI package conflicts are also reported as part of this check. They are reported separately from the configuration conflicts, in the same way as for a regular PyPI package compatibility check.

Console

To check for compatibility with Managed Airflow (Gen 3), use gcloud CLI or the Managed Airflow API. Google Cloud console supports only checks for PyPI package compatibility.

gcloud

To check for compatibility with Managed Airflow (Gen 3), run the environments check-upgrade command and specify a Managed Airflow (Gen 3) version as a target.

gcloud composer environments check-upgrade \
  ENVIRONMENT_NAME \
  --location LOCATION \
  --image-version COMPOSER_3_VERSION

Replace:

  • ENVIRONMENT_NAME with the name of the environment.
  • LOCATION with the region where the environment is located.
  • COMPOSER_3_VERSION with the Airflow version and build that you want to upgrade to, in the airflow-x.y.z-build.t format. You can use all version aliases.

Example:

gcloud composer environments check-upgrade example-environment \
  --location us-central1 \
  --image-version composer-3-airflow-2

Example conflict messages:

...
- message: Error validating property [core]unit_test_mode. Overriding [core]
unit_test_mode is not permitted.
type: BLOCKING
- message: Environment variables [GOOGLE_CLOUD_PROJECT] may not be overridden.
type: BLOCKING
- message: You have to specify Worker CPUs to be 0.5, 1.0 or multiples of 2.0.
type: BLOCKING
- message: The environment uses a default maintenance window, which is
different in Composer 2 and Composer 3. Unless set to a custom value, the
maintenance window will be changed to a new default after the upgrade.
type: NON_BLOCKING
...

As an alternative, you can run this check asynchronously. Use the --async argument to make an asynchronous call, then check the result with the gcloud composer operations describe command.

API

Construct an environments.checkUpgrade API request.

Specify the image version in the imageVersion field:

{
  "imageVersion": "COMPOSER_3_VERSION"
}

Replace COMPOSER_3_VERSION with the Airflow version and build that you want to upgrade to, in the airflow-x.y.z-build.t format. You can use all version aliases.

Upgrade your environment

To upgrade your environment to a later version of Managed Airflow or Airflow:

Console

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

    Go to Environments

  2. In the list of environments, click the name of your environment. The Environment details page opens.

  3. Go to the Environment configuration tab.

  4. Locate the Image version item and click Upgrade.

  5. From the Image version drop-down menu, select a Managed Airflow version that you want to upgrade to.

  6. Click Upgrade.

gcloud 

gcloud composer environments update \
  ENVIRONMENT_NAME \
  --location LOCATION \
  --image-version VERSION

Replace:

  • ENVIRONMENT_NAME with the name of the environment.
  • LOCATION with the region where the environment is located.
  • VERSION with the new Managed Airflow image version that you want to upgrade to, in the composer-a.b.c-airflow-x.y.z format. You can also use all version aliases.

For example:

gcloud composer environments update
  example-environment \
  --location us-central1 \
  --image-version composer-2.16.10-airflow-2.10.5

API

  1. Construct an environments.patch API request.

  2. In this request:

    1. In the updateMask parameter, specify the config.softwareConfig.imageVersion mask.

    2. In the request body, in the imageVersion field, specify a new version that you want to upgrade to.

For example:

// PATCH https://composer.googleapis.com/v1/projects/example-project/
// locations/us-central1/environments/example-environment?updateMask=
// config.softwareConfig.imageVersion

  {
    "config": {
      "softwareConfig": {
        "imageVersion": "composer-2.16.10-airflow-2.10.5"
      }
    }
  }

Terraform

The image_version field in the config.software_config block controls the Managed Airflow image of your environment. In this field, specify a new Managed Airflow image. 

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

  config {
    software_config {
      image_version = "VERSION"
    }
  }
}

Replace:

  • ENVIRONMENT_NAME with the name of the environment.
  • LOCATION with the region where the environment is located. the new Managed Airflow image version that you want to upgrade to, in the composer-a.b.c-airflow-x.y.z format. You can also use all version aliases.

Example:

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

  config {
    software_config {
      image_version = "composer-2.16.10-airflow-2.10.5"
    }
  }
}

What's next