Is there a recommended way of reading dbt model tags into Dagster for alerting? Context: • We load our DBT models as assets with

load_assets_from_dbt_project

• We set an

owner

meta tag within

sources.yml

for our models • We want to tag these owners using

make_slack_on_run_failure_sensor

. How can we access the tags via the

context

within a custom

text_fn

that’s passed to

make_slack_on_run_failure_sensor

? Our runs also typically contain multiple assets, potentially with multiple owners. Could we programmatically determine the owners of only the subset of models that failed in the run? (edited)

With

@dbt_assets

, since you have access to customize the compute function, you can invoke slack in the body of your assets definition. We showed this in our release livestream a while back: The basic code is here. You can imagine that you could just: 1. Get the

run_results_json

to get failed models 2. Look into the manifest object to get the

meta

associated with each failed model 3. Send a Slack message for each owner in (2)

@dbt_assets(manifest=dbt_manifest_path)
def jaffle_shop_dbt_assets(
    context: OpExecutionContext, dbt: DbtCliResource, slack: SlackResource
):
    dbt_build_invocation = dbt.cli(["build"], context=context)

    yield from dbt_build_invocation.stream()

    run_results_json = dbt_build_invocation.get_artifact("run_results.json")

    materialized_models_text = ""
    for result in run_results_json["results"]:
        if result["status"] == "success":
            materialized_models_text += (
                f"- `{result['unique_id']}`: {result['message']}\n"
            )

    slack.get_client().chat_postMessage(
        channel="team-nucleus-data-platform",
        text=(
            "The following dbt nodes have successfully materialized:\n"
            + materialized_models_text
        ),
    )

👑

Great, this has come up multiple times now, so we’ll add this as an example in our docs