https://dagster.io/ logo
Join Slack
Powered by
This message was deleted.
# ask-ai
s
This message was deleted.
s
In Dagster, the recommended way to access run information from a PostgreSQL database is through a 
DagsterInstance
object. This object provides a high-level API for interacting with the underlying storage system, which can include a PostgreSQL database when configured as such. The documentation provides an example of how to access the 
DagsterInstance
and use it to interact with the database:
Copy code
python from dagster import DagsterInstance  # Access the DagsterInstance instance = DagsterInstance.get()  # Use the instance to perform operations, such as retrieving runs runs = instance.get_runs()
This approach abstracts away the direct database interactions and provides a consistent interface for retrieving run information, regardless of the underlying storage system. If you are looking for a way to bypass the 
DagsterInstance
and interact directly with the PostgreSQL database, the documentation does not provide an explicit alternative method. Direct database access would require you to use SQL queries or a database client library like 
psycopg2
in Python, but this is outside the scope of the recommended practices in Dagster's documentation. Using the 
DagsterInstance
is the preferred method because it ensures that you are interacting with the database in a way that is consistent with Dagster's data models and abstractions. It also helps avoid potential issues that could arise from direct database manipulation, such as schema changes or data integrity concerns. If you have a specific use case that requires direct database access, you would need to handle the database connection and queries yourself, but this is not covered by the documentation and would not be considered a best practice within the context of Dagster's design.
If you are unable to use 
DagsterInstance.get()
in your environment, another way to interact with Dagster's underlying data is through the Dagster GraphQL API. The documentation mentions the 
DagsterGraphQLClient
, which provides a Python interface to interact with the GraphQL API without needing to construct GraphQL queries manually. Here's an example of how you might use the 
DagsterGraphQLClient
to get the status of a job run:
Copy code
python
from dagster_graphql import DagsterGraphQLClient, DagsterGraphQLClientError
from dagster import DagsterRunStatus

# Instantiate the client with your Dagster instance's host and port
client = DagsterGraphQLClient("localhost", port_number=3000)

# Use the client to get the status of a job run
try:
    status: DagsterRunStatus = client.get_run_status(RUN_ID)
    if status == DagsterRunStatus.SUCCESS:
        # Handle success
    else:
        # Handle other statuses
except DagsterGraphQLClientError as exc:
    # Handle exception
This client provides a simplified interface for certain operations, such as submitting a job run, getting the status of a run, reloading a repository location, and shutting down a repository location. However, it does not provide direct access to the full set of run data that you might get from the 
DagsterInstance
API. If the 
DagsterGraphQLClient
does not provide the methods you need, and you cannot use 
DagsterInstance.get()
, you may need to consider other options, such as: 1. Direct database access: As mentioned earlier, this would involve using a PostgreSQL client library to execute SQL queries against the Dagster run storage tables. This approach requires a good understanding of Dagster's database schema and is not recommended because it bypasses Dagster's abstractions and could lead to issues if the schema changes. 2. Custom API: If you have control over the Dagster deployment, you could potentially expose a custom API endpoint that provides the necessary run data. This would involve extending the Dagster webserver or creating a separate service that interacts with the Dagster database. 3. Dagster CLI: For some use cases, you might be able to use the Dagster CLI to retrieve run information, which can then be processed further in your environment. Please note that these alternatives are not covered in detail by the documentation and may require a deeper integration effort on your part. They also come with the caveats mentioned above regarding direct database access and maintaining compatibility with Dagster's data models. How to access the run event logs of all ... button How do I access a 
DagsterInstance
from... button How to access the currently running Dags... button grpc server crashing with ReloadCode Not... button Thank you for your feedback!
Open in Slack
PreviousNext
Powered by