Schedular

DAGSTER INTEVIEW QUESTIONS

4.1 (1910)11,110 learners20 lessons40m

Curriculum

Topic

What is Dagster and how does its asset-centric model differ from a task-centric orchestrator like Airflow_1:56
What is a software-defined asset, and why does Dagster treat the data (the asset) rather than the task as the core unit of orchestration_1:58
Explain the difference between assets, ops, jobs, and graphs in Dagster, and when each still matters.1:50
You're migrating an Airflow DAG of tasks (_extract_, _transform_, _load_) to Dagster. How would you rethink it in terms of assets, and what do you gain_1:53
Your pipeline produces 200 tables and stakeholders constantly ask _where did this number come from__ How does Dagster's asset lineage help, and how would you set it up_2:04
One asset depends on another asset's data but you don't want to pass the data in memory between them. How do dependencies and IO managers handle this_1:53
You have an existing dbt project with hundreds of models. How would you bring it into Dagster so each dbt model becomes an observable asset_2:14
Some data lives in an external system you don't control (e.g., a table populated by Fivetran). How would you represent it in Dagster so downstream assets can depend on it_1:54
You ingest one file per day and need to process each day independently and reprocess specific past days on demand. How would you model this with partitions_1:53
A bug corrupted the last 30 days of a daily-partitioned asset. How would you backfill just those partitions without recomputing all of history_2:04
Your partitions are defined by both date and region (a multi-dimensional grid). How would you set this up and what are the operational caveats_2:17
An upstream partitioned asset only has some partitions filled in. How do you make downstream partitions materialize automatically as soon as their upstream dependencies are ready_2:08
You need an asset to refresh every day at 6 AM, but only after its upstream source has landed. How would you choose between a schedule, a sensor, and declarative automation_2:00
A pipeline should run whenever a new file appears in an S3 bucket, not on a fixed clock. How would you implement this_2:16
You want assets to rematerialize automatically whenever their upstream data changes — without writing cron schedules at all. How does declarative automation achieve this_2:00
Two teams both depend on the same upstream asset and were each running their own redundant schedules to rebuild it. How does Dagster's model eliminate that wasted computation_2:02
You need to fail a pipeline (or block downstream assets) when an asset's data violates a quality rule, like null primary keys. How would you implement this in Dagster_1:56
A downstream asset was silently built on bad data because a quality check was skipped rather than run. How do blocking vs. non-blocking asset checks change this behavior_1:52
You want to unit-test your pipeline logic locally without hitting production databases. How does Dagster's resource_config system let you swap implementations for tests_1:45
A run failed halfway through a multi-asset job. How would you re-execute only the failed and downstream assets rather than the whole job_2:00