Pyspark code running twice causing a LIVY status = DEAD

Hello i am running the following cell (all packages has been imported and parameters set in a previous cells, and the code is running)

df = spark.sql(f"select * from LH.STOCK_Price where YEAR = {parameter_year} and MONTH <= {parameter_month}")

# Filter the DataFrame to include only rows where the year in TIMESTAMP is parameter
#df = df.filter((col("YEAR") == parameter_year) & (col("MONTH") <= parameter_month))

#df.cache()

# Ensure the data is partitioned appropriately
df = df.repartition("TYPE", "Vendor", col("TIMESTAMP").cast("date"))

# Define a custom function to count distinct IDs
distinct_count = F.expr("count(distinct ID)").alias("ID_DISTINCT_COUNT")

# Group by TYPE, Vendor, and TIMESTAMP (cast to date)
result = df.groupBy(
    col("TYPE"),
    col("Vendor"),
    col("TIMESTAMP").cast("date").alias("DATE")
).agg(
    count(col("ID")).alias("ID_COUNT"),
    distinct_count
)

result.cache()

# Write the result to a Delta table with partitioning
result.write.format("delta").mode("append").partitionBy("DATE").saveAsTable("LH.Count_Date_Wise_STOCKS")

The aim of my code is to read a STOCK_Price delta table from a lakehouse for a specific year month into a dataframe df, then group by different columns and count the ID and their distinct count. I am caching the result and then writing it to an aggregate delta table (Count_Date_Wise_Stocks), but i feel that part of my code is running twice as i can see from the spark diagnostic below the cell:

The second run is almost producing a session error with an error code LIVY state = Dead.

Some statistics about my tables:
STOCK_Price : around 700 Billion Rows
df : about 25 Billlions Rows
Count_Date_Wise_STOCKS: around 10,000 rows only.
the execution of the cell is taking sometimes up to 10 hours for 6 month of data and returning a Count_Date_Wise_STOCKS of around 4000 rows.

Questions: What part of my code is causing the second run? how can i avoid that if it is possible?
Regards