PySparkBeginnerBroken PySpark FixFree

The Cache Everything Trap

A notebook-derived pipeline caches several large DataFrames because developers wanted to 'speed things up'.

Scenario context

Production jobs show higher memory pressure, more spills, and worse overall runtime than before caching was added.

Business requirement

Fix the PySpark code so the pipeline is correct, scalable, and safe to rerun.

Schema

DataFrames depend on the scenario. Assume large production-scale inputs, skewed keys, retries, and partitioned lake storage.

Sample input

Use the code comments and logs to infer the input shape. Focus on the production failure mode, not local toy execution.

Broken logic / code

# Broken: caching every intermediate DataFrame consumes executor memory.
raw = spark.read.parquet(raw_path).cache()
filtered = raw.filter("event_date = '${run_date}'").cache()
joined = filtered.join(dim_customers, 'customer_id', 'left').cache()
result = joined.groupBy('customer_id').count().cache()
result.write.mode('overwrite').parquet(output_path)

Logs / error

[Spark] The Cache Everything Trap
Stage progress: most tasks finished, one or more tasks are long-running.
Metrics to inspect: shuffle read, spill, skew ratio, executor lost count, file count.
Production jobs show higher memory pressure, more spills, and worse overall runtime than before
caching was added.

Expected output / expected logic

Corrected PySpark code or approach should reduce the failure mode, preserve correctness, and include validation/monitoring.

Your attempt

Write the corrected PySpark approach

Think before revealing the answer. A partial but honest attempt is better practice than reading the model solution first.

Saved locally

Interview-style explanation

Now explain your solution as if you are in an interview: symptom, root cause, fix, edge cases, trade-offs, monitoring, and prevention.