The Storage tab (shown in Caching and Persistence Levels) lists every cached DataFrame with its level, partitions cached, and size in memory and on disk. The Executors tab summarizes each executor, here the single local-mode driver.
"Storage Memory" reads 36.5 MiB used of 434.4 MiB: the cached order lines against the memory that the unified pool can lend to storage. With the default 1 GB driver, that pool is (1,024 MiB minus 300 MiB reserved) times spark.memory.fraction 0.6, exactly 434.4 MiB (The Executor Memory Model explains the model). Watch "Failed Tasks" and "Dead" executors for crashes and lost containers, GC time relative to task time for memory pressure, and per-executor input and shuffle columns for an executor that does far more than its peers. The Thread Dump link shows what a stuck task is doing; it is how the hang in Pandas UDFs (Vectorized UDFs) was traced to tasks waiting on a Python worker.