pbzip2 produces a separate bzip2 stream for every compressed block, which can't be read properly by many popular decoders, usually ones that are based around the bzip2 high-level interface [0]. There's a warning about this in Python 2's documentation, which is a better offering than most [1].
Many decoders simply stop reading after decompressing the first 900k without so much as a peep that something that may astonish you has happened.
Many decoders simply stop reading after decompressing the first 900k without so much as a peep that something that may astonish you has happened.