dagster-sensors-to-orchestra

Dagster Sensors -> Orchestra Sensors Block

Overview

Dagster sensors are functions evaluated on a tick interval that inspect external state and yield RunRequest(...) to launch a run when a condition is met (or SkipReason otherwise). In Orchestra, sensors are pipeline triggers — they live in the sensors: block at the pipeline root (not inside pipeline:). When all of a sensor's checks pass, the pipeline run is triggered automatically.

Key difference: Dagster sensors run arbitrary Python every tick. Orchestra sensors use a cron window + polling interval model with declarative checks drawn from SensorChecksEnum.

SensorModel Structure

sensors:
  <sensor-id>:
    name: My Sensor
    cron: '0 8 * * ? *'               # when the check window opens
    timezone: UTC
    timeout_mins: 60                 # max 7200; must be < cron interval
    frequency_secs: 60               # polling interval (60-600)
    exclude: []
    run_inputs: {}

    checks:
      <check-id>:
        integration: SNOWFLAKE
        sensor_type: SNOWFLAKE_QUERY
        connection: my_snowflake_12345
        parameters:
          query: "SELECT COUNT(*) FROM daily_files WHERE date = CURRENT_DATE"
        map_outputs:
          file_count: "result"

    alerts:
      - name: sensor-timed-out
        statuses: [FAILED]
        destinations:
          - integration: SLACK
            destination: '#data-alerts'

Valid SensorChecksEnum Values