Save and load environment snapshots

Cloud Composer 3 | Cloud Composer 2 | Cloud Composer 1

This page explains how to save and load the state of your environment using environment snapshots.

You can configure your environment to automatically save snapshots. For more information, see Configure scheduled snapshots.

About environment snapshots

Environment snapshots store the state of your environment. You can save and load environment snapshots on demand.

You can use snapshots to:

• Migrate your environments to a later major version of Cloud Composer or Airflow.

How snapshots are stored

An environment snapshot is a set of files that describe the state of your environment and store the backup of the environment data.

You can create multiple snapshots of your environment. Environment snapshots are non-incremental. You can use any snapshot independently of other snapshots.

Cloud Composer does not delete snapshots when you delete your environment.

By default, Cloud Composer stores snapshots in the snapshots/ folder in your environment's bucket. You can also specify a custom location when you create a snapshot.

While the Cloud Storage UI displays data as folders for convenience, buckets work in a flat-file structure. This doesn't change how you work with saving and loading snapshots in the Google Cloud console, but keep this in mind when you perform operations directly on the files in your bucket. For more information, see Object namespace.

Security considerations for snapshots

To mitigate this security risk, you can store sensitive information that is used by Airflow DAGs, such as keys or passwords, in Secret Manager. For more information, see Configure Secret Manager for your environment.

Make sure to check security permissions for your environment's bucket. If you store environment snapshots in a custom bucket, make sure that access permissions for it are configured properly in your project. When assigning permissions, make sure that environment's service account has enough permissions to save and load snapshots from the bucket.

What data is saved in snapshots

Cloud Composer saves the following data in snapshots:

• Airflow configuration overrides.
• Environment variables.
• List of custom PyPI packages, as requirements.
• A backup of the Airflow database, including states of executed tasks, and DAG runs history.
• A backup of the /dags, /data, and /plugins folders from the environment's bucket.
• Environment's fernet key.
• Other information about the environment's configuration, such as environment's scale and performance parameters. Cloud Composer does not use this information when it loads snapshots.

What data is loaded from snapshots

Cloud Composer loads the following data from snapshots:

• Airflow configuration overrides.
• Environment variables.

• Custom PyPI packages (unless you choose to skip installing them).

• The contents of the Airflow database, including states of executed tasks, and DAG runs history.

• Contents of the /dags, /data, and /plugins folders from the snapshot are loaded into the environment's bucket.

• The fernet key from the snapshot is used to re-encrypt the data from the snapshot with the environment's own fernet key. The fernet key of the environment remains unchanged.

Although Cloud Composer stores some information about the environment's configuration in snapshots, it is not used when loading snapshots. The following parameters of your environment do not change when you load a snapshot:

• Environment configuration, such as environment scale and performance parameters.
• Environment's networking configuration.
• Contents of the environment's bucket outside of the /dags, /data, and /plugins folders.
• Environment labels.

Any settings that you applied in Cloud Composer infrastructure without using Cloud Composer API might be lost when you load a snapshot.

About partially completed operations

When you load a snapshot, the operation can be successful, failed, or partially completed:

• Successful operations load all data from the snapshot.
• Failed operations do not introduce any changes.
• Partially completed operations load a subset of data from the snapshot. Such operations are reported as failed, but the error message indicates what data was successfully loaded. For example, if PyPI packages are installed, but Airflow configuration option overrides did not succeed, the error message indicates this.

For a partially completed operation, you can try to load the same snapshot again. Cloud Composer skips steps that were successful on the previous attempt. For example, if an operation failed on a timeout, but the database was successfully loaded, then the next attempt does not load the database again.

Before you begin

• In Cloud Composer 1 you can only save snapshots, but not load them. You can load snapshots from a Cloud Composer 1 environment to Cloud Composer 2 environments.

• Snapshots are supported in Cloud Composer 2 version 2.0.9 and later. Cloud Composer 1 supports saving environment snapshots in 1.18.5 and later versions.

• Snapshots do not create an environment. If you want to load a snapshot from an environment to a different environment, you first need to create a new environment and then load the snapshot to it.

• You cannot load snapshots to environments that are in the error state. It is not possible to fix such environments by loading a snapshot. You can still load an existing snapshot to a new environment.

• The maximum size of the Airflow database that supports snapshots is 20 GB. If your environment's database takes more than 20 GB, reduce the size of the Airflow database before saving a snapshot.

• The total number of objects in the /dags, /plugins and /data folders in the environment's bucket must be less than 100,000 to create snapshots.

• If you use the XCom mechanism to transfer files, make sure that you [use it according to Airflow's guidelines][ext-airflow-xcom-backend]. 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.

• The maximum length of an object's name in the environment's bucket must be no longer that 1024 symbols. Otherwise, snapshots cannot be created.

• If you save snapshots in a location outside your environment's bucket, the service account of your environment must have read and write permissions for the specified location. For example, the Storage Object Admin role has such permissions. You can apply it to a project or to a specific bucket.

• Creating and storing snapshots produces additional costs, which are related to Cloud Storage. For more information, see Pricing.

Save an environment snapshot

Cloud Composer saves environment snapshots in a subfolder, relative to the folder that you specify. The folder name contains the project ID, the environment's location, the environment's name, and the timestamp when the snapshot was saved. For example: /snapshots/example-project_us-central1_example-environment_2022-01-05T18-59-00.

gcloud composer environments snapshots save \
  ENVIRONMENT_NAME \
  --location LOCATION \
  --snapshot-location "SNAPSHOTS_FOLDER"

gcloud composer environments snapshots save \
  example-environment \
  --location us-central1

gcloud composer environments snapshots save \
  example-environment \
  --location us-central1 \
  --snapshot-location "gs://example-bucket/environment_snapshots"

{
  "snapshotLocation": "SNAPSHOTS_FOLDER"
}

// POST https://composer.googleapis.com/v1/projects/example-project/
// locations/us-central1/environments/example-environment:saveSnapshot

{
  "snapshotLocation": "gs://us-central1-example-916807e1-bucket/snapshots"
}

Load an environment snapshot

In Cloud Composer 1 you can only save snapshots, but not load them. You can load snapshots from a Cloud Composer 1 environment to Cloud Composer 2 environments. For example, when you migrate your environments to Cloud Composer 2.

What's next

• Configure scheduled snapshots
• Create environments
• Access control