Prepare Airflow interview answers on DAG dependencies, task retries, intervals, backfills and data quality with a repeated-daily-job scenario. These original practice questions connect a concept to a decision and a failure case. They are preparation exercises, not leaked employer questions. State your assumptions before proposing an implementation.
What should a DAG describe?
It describes task dependencies and the workflow's execution structure. Keep orchestration distinct from the business meaning of the data. Explain the output each task produces and what its dependents require. A dependency arrow establishes order, but it does not guarantee that the upstream result is complete or correct. Add appropriate validation at the boundary where downstream work relies on it.
How do you make a task safe to retry?
Define the logical input and output so repeating the same work does not create unintended duplicates or inconsistent results. Use a stable interval or batch identity and an appropriate write policy. Avoid relying on the current wall-clock date for every rerun. Determine whether a previous attempt committed before it failed and reconcile that state before blindly applying the effect again.
Why do data intervals matter?
A scheduled run often represents a defined slice of data rather than whatever data exists at the instant it happens to execute. State which timestamps select records and how late arrivals are handled. A delayed run and a backfill should still process the intended interval. Explain the business date separately from execution time so yesterday's recovery does not accidentally load today's records.
What makes a backfill risky?
Historical runs can place unusual load on dependencies and repeat writes or external effects. Check interval selection, target partition behavior and concurrency limits before starting. Explain whether existing output is replaced, merged or retained. Test one interval first with reconciled results. A workflow that succeeds for the latest day can still mishandle historical schema differences or corrected inputs.
How should tasks exchange information?
Use the supported mechanism for small control information and suitable storage for substantial datasets. Make the data location and identity explicit so another worker can access it. Do not assume that a local file created by one task is available to a task running elsewhere. Explain access, retention and cleanup, and validate that the referenced data belongs to the intended run.
Worked interview scenario
Original job: a daily task appends sales totals using the current date at execution time. It fails after writing, then a retry after midnight writes the wrong day's label and duplicates a total. Define a stable data interval and a write policy for that interval.
Use the interval's source data and produce a deterministic partition or keyed result according to the reporting contract. Test an interrupted attempt, a delayed run and a historical backfill. Add a check comparing source and output totals before the report is considered ready. If the task also sends an external notification, define that separate side effect's repeat policy rather than assuming a correct table write covers it.
Practice exercise
Draw a three-task workflow for extraction, transformation and validation. State each output and its interval identity. Rehearse the midnight failure case and describe a safe one-day backfill. Finish by explaining how another worker finds the data without depending on the first worker's local filesystem.
Review your explanation
Use Cluegent during practice to challenge your own draft. Ask for a follow-up about the scenario's weakest assumption, answer it without suggestions, then check your reasoning against the official reference. Review current plans before subscribing. Follow the employer's tool policy in the actual interview.
Sources checked
These official references support the guide. Product details and technical documentation can change; check the linked source for current information.
Where Cluegent helps
Cluegent supports permitted live workflows with transcript context, typed prompts, screenshot-aware answers, resume context, custom response behavior, quick action buttons, and a private desktop overlay. It is most useful when you already understand the subject and need help staying structured under pressure.
Frequently asked questions
Should every retry use today's date?
Use the logical interval or batch the run represents. Execution time can change during delays and retries.
Does task order guarantee data quality?
No. Validate the output's identity, completeness and business totals before downstream reporting relies on it.