I’m trying to create a partitioned table in Clickhouse with:

CREATE TABLE IF NOT EXISTS new_table (
  logging_day Date,
  sensor_name String,
  ts       DateTime
  ts_hour  DateTime MATERIALIZED toStartOfHour(ts),
)
ENGINE=MergeTree()
PARTITION BY logging_day
ORDER BY (sensor_name, ts_hour, ts)
PRIMARY KEY (sensor_name, ts_hour);

The table is partitioned by logging_day. From query performance perspective, does it still need be added to the ORDER BY or PRIMARY KEY lists like:

ORDER BY (logging_day, sensor_name, ts_hour, ts)
PRIMARY KEY (logging_day, sensor_name, ts_hour);

Which way will make the WHERE filtered queries faster, with or without the partition key in the lists? Thanks.