I’ve never played with HDFS snapshots, just going from this:
https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HdfsSnapshots.html
Looks like the snapshot is just taking a snapshot of the metadata. Any file that is deleted will be removed from the active meta, but the individual blocks wouldn’t be actually deleted since they are still referenced by the snapshot. I’m guessing at this point the blocks would show up as non-dfs space used.