Setting up the Alert Production Database for ap_pipe

In its default configuration, the Alert Production Pipeline, as represented by pipelines/_ingredients/ApPipe.yaml, relies on a database to save and load DIASources and DIAObjects. When running as part of the operational system, this database will be provided externally. However, during testing and development, developers can run apdb-cli to set up their own database. This page provides an overview of how to integrate the AP pipeline with apdb-cli.

Configuring the database

The database is configured using apdb-cli, as described in the dax_apdb documentation. If you are creating your own database, apdb-cli outputs a config file encoding the database settings. If you are using a database set up by someone else, they should have provided you with a config file or its label in the APDB index (see lsst.dax.apdb.Apdb.from_uri() for details).

Once you have a config file or label, you must pass it to the pipeline as the parameters:apdb_config config field (see --config). Note that this parameter has no default; a value must be provided by the user.

Additionally, the default set of observed bands allowed to be used in the pipeline are set by the columns available in the Apdb schema specified by schema_file. Should the user wish to use the pipeline on data containing bands not in the ugrizy system, they must add the appropriate columns to the Apdb schema and add the bands to the validBands config in DiaPipelineConfig.


To create an SQLite database from scratch, run the following (see the dax_apdb documentation for an explanation of apdb-cli, and Running the AP pipeline for an explanation of pipetask):

apdb-cli create-sql sqlite:///databases/apdb.db
pipetask run -p ApPipe.yaml -c differencer:coaddName=dcr -b repo -o myrun

The argument to apdb-cli specifies the name of the created configuration file that will contain a serialized ApdbConfig for the new database.

A Postgres database can be set up and used with the following:

apdb-cli create-sql --namespace='my_apdb_name' 'postgresql://'
pipetask run -p ApPipe.yaml -c -d "my_data_query" -b repo -i my/input/collection -o my/output/collection

If a pre-existing database is registered in the dax_apdb index, this becomes:

pipetask run -p ApPipe.yaml -c parameters:apdb_config=label:db_name -d "my_data_query" -b repo -i my/input/collection -o my/output/collection

A Postgres database can be set up and used within bps yaml files by adding this to a submit yaml:

extraQgraphOptions: "-c parameters:apdb_config=/path/to/"
apdb-cli create-sql --namespace='my_apdb_name' 'postgresql://'

Note that apdb-cli must be run prior to submitting this bps yaml, and the path to the resulting config file ( in this example) passed in extraQgraphOptions.

Further reading