Nightly Data Builds#

The complete ETL and tests are run each night on a Google Batch-managed VM to ensure that any new changes merged into main are fully tested. These complete builds also enable continuous deployment of PUDL’s data outputs. If no changes have been merged into main since the last time the builds ran, the builds are skipped.

The builds are kicked off by the build-deploy-pudl GitHub Action, which builds and pushes a Docker image with PUDL installed to Docker Hub and then launches a Google Batch job using that image. Inside the container, builds/pudl_batch.sh runs the ETL and tests, saves the raw build outputs to gs://builds.catalyst.coop, and if successful publishes the distributable outputs to our public cloud buckets.

Breaking the Builds#

The nightly data builds based on the main branch are our comprehensive integration tests. When they pass, we consider the results fit for public consumption. The builds are expected to pass. If they don’t then someone needs to take responsibility for getting them working again with some urgency.

Because of how long the full build and tests take, we don’t typically run them individually before merging every PR into main. However, if you’ve added a new year of data or made changes that are likely to affect the full ETL or data validations, it is often worth running a full local build that resembles the nightly builds:

$ pixi run pudl-with-ferc-to-sqlite
$ pixi run pytest-nightly

For local development, we recommend pixi run pudl-with-ferc-to-sqlite rather than pixi run pudl-with-ferc-to-sqlite-nightly. The nightly ETL task uses the same asset graph and datasets, but it is tuned for the higher-resource nightly build environment and enables more verbose logging.

If your PR causes the build to fail, you are probably the best person to fix the problem, since you already have context on all of the changes that went into it.

Having multiple PRs merged into main simultaneously when the builds are breaking makes it ambiguous where the problem is coming from, makes debugging harder, and diffuses responsibility for the breakage across several people, so it’s important to fix the breakage quickly. In some cases we may delay merging additional PRs into main if the builds are failing to avoid ambiguity and facilitate debugging.

Therefore, we’ve adopted the following etiquette regarding build breakage: On the morning after you merge a PR into main, you should check whether the nightly builds succeeded by looking in the pudl-deployments Slack channel (which all team members should be subscribed to). If the builds failed, look at the logging output (which is included as an attachment to the notification) and figure out what kind of failure occurred:

  • If the failure is due to your changes, then you are responsible for fixing the problem and making a new PR to main that resolves it, and it should be a high priority. If you’re stumped, ask for help!

  • If the failure is due to an infrastructural issue like the build server running out of memory and the build process getting killed, then you need to notify the member who is in charge of managing the builds (Currently @zaneselvans), and hand off responsibility for debugging and fixing the issue.

  • If the failure is the result of a transient problem outside of our control like a network connection failing, then wait until the next morning and repeat the above process. If the “transient” problem persists, bring it up with the person managing the builds.

The GitHub Action#

The build-deploy-pudl GitHub action contains the main coordination logic for the Nightly Data Builds. The action is triggered every night and when new versioned release tags are pushed to the PUDL repository. This way, new data outputs are automatically updated for releases, and PUDL’s code and data are tested every night.

The gcloud command in build-deploy-pudl requires certain Google Cloud Platform (GCP) permissions to start and update the Google Batch VM. We use Workflow Identity Federation to authenticate the GitHub Action with GCP in the GitHub Action workflow.

Deployment Action#

The experimental deploy-pudl action separates deployment from the build process. This action takes a git tag that has already been built as an input and will find the corresponding build outputs and determine the deployment type (stable or nightly) from the tag. It will then upload outputs from the build to GCS and S3, update the git branch associated with the deployment type, and trigger a zenodo release. This action can also take an optional staging flag will upload outputs to a dedicated staging area, and will not update the git branch or trigger a Zenodo release.

Eventually, the deployment functionality will be removed from the build-deploy-pudl action and it will instead trigger this action at the end of a successful build.

Google Compute Engine#

We use ephemeral VMs created with Google Batch to run the nightly builds. Once the build has finished – successfully or not – the VM shuts itself down. The build VMs use the e2-highmem-8 machine type (8 CPUs and 64GB of RAM) to accommodate the PUDL ETL’s memory-intensive steps. Currently, these VMs do not have swap space enabled, so if they run out of memory, the build will immediately terminate.

The deploy-pudl-vm-service-account service account has permissions to:

  1. Write logs to Cloud Logging.

  2. Start and stop the VM so the container can shut the instance off when the ETL is complete, so Catalyst does not incur unnecessary charges.

  3. Bill the catalyst-cooperative-pudl project for egress fees from accessing the zenodo-cache.catalyst.coop bucket. Note: The catalyst-cooperative-pudl won’t be charged anything because the data stays within Google’s network.

  4. Write logs and build outputs to the gs://builds.catalyst.coop, gs://pudl.catalyst.coop and s3://pudl.catalyst.coop buckets. Egress and storage costs for the S3 bucket are covered by Amazon Web Services’s Open Data Sponsorship Program.

Build outputs and logs are saved to the gs://builds.catalyst.coop bucket so you can access them later. Build logs and outputs are retained for 30 days and then deleted automatically.

Docker#

The Docker image the VMs pull installs the PUDL pixi environment. The VMs are configured to run the builds/pudl_batch.sh script. This script:

  1. Notifies the pudl-deployments Slack channel that a deployment has started. Note: if the container is manually stopped, slack will not be notified.

  2. Runs pixi run pudl-with-ferc-to-sqlite-nightly.

  3. Runs pixi run pytest-unit-nightly, pixi run pytest-integration-nightly, and pixi run pytest-data-validation-nightly as separate stages.

  4. Copies the outputs and logs to a directory in the gs://builds.catalyst.coop bucket. The directory is named using the git SHA of the commit that launched the build.

  5. Copies the outputs to the gs://pudl.catalyst.coop and s3://pudl.catalyst.coop buckets if the ETL and test suite run successfully.

  6. Notifies the pudl-deployments Slack channel with the final build status, including per-stage status and durations.

The pudl_batch.sh script is only intended to run on a Google Batch VM with adequate permissions.

The nightly ETL task, nightly pytest tasks, and the nightly build script all share the same DG_NIGHTLY_CONFIG environment variable, which points at src/pudl/package_data/settings/dg_nightly.yml relative to the repository root. Using a repo-relative path avoids a second hard-coded config path for the container and keeps the nightly build behavior centralized in pixi while still allowing local runs to reuse the nightly pytest commands.

How to access the nightly build outputs from AWS#

You can download the outputs from a successful nightly build data directly from the s3://pudl.catalyst.coop bucket using the gcloud storage CLI, which can access both GCS and S3 storage buckets.

gcloud storage ls s3://pudl.catalyst.coop

You should see a list of directories with version names:

s3://pudl.catalyst.coop/nightly/
s3://pudl.catalyst.coop/stable/
s3://pudl.catalyst.coop/v2025.10.0/
s3://pudl.catalyst.coop/v2025.11.0/
s3://pudl.catalyst.coop/v2025.12.1/
s3://pudl.catalyst.coop/v2025.2.0/
s3://pudl.catalyst.coop/v2025.5.0/
s3://pudl.catalyst.coop/v2025.7.0/
s3://pudl.catalyst.coop/v2025.8.0/
s3://pudl.catalyst.coop/v2025.9.0/
s3://pudl.catalyst.coop/v2025.9.1/
...

Warning

If you download the files locally then you’ll be responsible for updating them, making sure you have the right version, putting them in the right place on your computer, etc.

To copy these files directly to your computer you can use the gcloud storage cp command, which behaves very much like the Unix cp command:

gcloud storage cp s3://pudl.catalyst.coop/nightly/pudl.sqlite.zip ./

If you wanted to download all of the build outputs (more than 25GB!) you can use a recursive copy:

gcloud storage cp --recursive s3://pudl.catalyst.coop/nightly/ ./

How to access the nightly build outputs and logs (for the Catalyst team only)#

Sometimes it is helpful to download the logs and data outputs of nightly builds when debugging failures. To do this you’ll need to set up the Google Cloud Software Development Kit (SDK). It is installed as part of the PUDL pixi environment.

To authenticate with Google Cloud Platform (GCP) you’ll need to run the following:

gcloud auth login

Initialize the gcloud command line interface and select the catalyst-cooperative-pudl project.

If it asks you whether you want to “re-initialize this configuration with new settings” say yes.

gcloud init

Finally, use gcloud to establish application default credentials; this will allow the project to be used for requester pays access through applications:

gcloud auth application-default login

Tip

If you’ve done all this and you are still getting “ERROR: (gcloud.storage.hash) HTTPError 400: Bucket is a requester pays bucket but no user project provided.” errors below, try:

gcloud config set billing/quota_project catalyst-cooperative-pudl

To test whether your GCP account is set up correctly and authenticated you can run the following command to list the contents of the cloud storage bucket containing the PUDL data. This doesn’t actually download any data, but will show you the versions that are available:

gcloud storage ls --long --readable-sizes gs://builds.catalyst.coop

You should see a list of directories with build IDs that have a naming convention: <YYYY-MM-DD-HHMM>-<short git commit SHA>-<git branch>.

To see what the outputs are for a given nightly build, you can use gcloud storage like this:

 gcloud storage ls --long --readable-sizes gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main

   6.60MiB  2024-11-15T13:28:20Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/2024-11-15-0603-60f488239-main.log
  804.57MiB  2024-11-15T12:40:35Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/censusdp1tract.sqlite
  759.32MiB  2024-11-15T12:41:01Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc1_dbf.sqlite
    1.19GiB  2024-11-15T12:41:12Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc1_xbrl.sqlite
    2.16MiB  2024-11-15T12:39:23Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc1_xbrl_datapackage.json
    6.95MiB  2024-11-15T12:39:23Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc1_xbrl_taxonomy_metadata.json
  282.71MiB  2024-11-15T12:40:40Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc2_dbf.sqlite
  127.39MiB  2024-11-15T12:39:59Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc2_xbrl.sqlite
    2.46MiB  2024-11-15T12:40:54Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc2_xbrl_datapackage.json
    6.82MiB  2024-11-15T12:40:48Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc2_xbrl_taxonomy_metadata.json
    8.25MiB  2024-11-15T12:39:22Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc60_dbf.sqlite
   27.89MiB  2024-11-15T12:39:24Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc60_xbrl.sqlite
  942.19kiB  2024-11-15T12:39:22Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc60_xbrl_datapackage.json
    1.77MiB  2024-11-15T12:39:22Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc60_xbrl_taxonomy_metadata.json
  153.72MiB  2024-11-15T12:41:03Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc6_dbf.sqlite
   90.51MiB  2024-11-15T12:41:09Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc6_xbrl.sqlite
    1.32MiB  2024-11-15T12:40:47Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc6_xbrl_datapackage.json
    2.74MiB  2024-11-15T12:39:22Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc6_xbrl_taxonomy_metadata.json
    1.38GiB  2024-11-15T12:41:06Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc714_xbrl.sqlite
   83.39kiB  2024-11-15T12:40:46Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc714_xbrl_datapackage.json
  187.86kiB  2024-11-15T12:40:46Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/ferc714_xbrl_taxonomy_metadata.json
   15.06GiB  2024-11-15T12:42:17Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/pudl.sqlite
         0B  2024-11-15T12:39:22Z  gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/success
                                   gs://builds.catalyst.coop/2024-11-15-0603-60f488239-main/parquet/
TOTAL: 23 objects, 21331056422 bytes (19.87GiB)

If you want to copy these files down directly to your computer, you can use the gcloud storage cp command, which behaves very much like the Unix cp command:

gcloud storage cp gs://builds.catalyst.coop/<build ID>/pudl.sqlite ./

If you need to download all of the build outputs (~20GB!) you can do a recursive copy of the whole directory hierarchy (note that this will incur egress charges):

gcloud storage cp --recursive gs://builds.catalyst.coop/<build ID>/ ./

For more background on gcloud storage see the quickstart guide or check out the CLI documentation with:

gcloud storage --help