Sam
07/26/2023, 3:03 PMweather_data_df
has the latest partition materialized
• latest_weather_data_df
is set to eager
• latest_weather_data_df
has never been materialized before
I expect Materialization is missing
to be checked. Though that is not the case after 29 evaluations.
I suspect this is due to usage of LastPartitionMapping
to clear the partition.
For completeness, the previous 2 skips were due to Waiting on upstream data
Sam
07/26/2023, 3:08 PMdomain_model
• daily partitioned weather_data_df
I want domain_model
to build with the latest materialized partition of weather_data_df
I believe this is currently not possible with partition mapping.
What I end up doing is creating a latest_weather_data_df
using LastPartitionMapping
that eagerly updates.Daniel Gafni
07/26/2023, 3:08 PMmax_materializations_per_minute
to a higher value to achieve the backfill effectSam
07/26/2023, 3:17 PMweather_data
, but I wasn’t able to get the weather_data_df
to trigger any Materialization condition.
Given
• weather_data
’s first partition is there
• `weather_data_df`’s first partition has not materialized before
• I waited for the Evaluation History
count to increment
I expect Materialization is missing
to be true, but that is not the case.Sam
07/26/2023, 3:18 PMMaterialization is missing
not triggered for backfill case, and
2. Why is Materialization is missing
not triggered from the most recent partition for latest_weather_data_df
, which I suspect is related to LastPartitionMapping
usageSam
07/26/2023, 4:00 PMLastPartitionMapping
or lost it through aggressive ctrl-z-ing.
Setting LastPartitionMapping
made it work with only the latest partition available as expected.
Without specifying LastPartitionMapping
in the ins
, the default behavior is to pass all partitions in a dict. So I can appreciate why it didn’t start, but hard to figure that out by staring.
Trying to figure out 1 now.