How do I make sure gzipped parquet files are the correct size on hdfs?

Question

I\'m writing thousands of data frames as gzip parquet files with a target size of 512mb. However, hitting 512mb is hard. With a number of heuristics, i.e. compression ratios