Custom compressioncodec on Spark -
has developed java-based custom compressioncodec apache spark (i.e. org.apache.spark.io.compressioncodec)?
unlike older versions, requires implementation of 2 methods in class: compressedinputstream , compressedoutputstream
i'm working on this, , works smaller data sizes, shows "executor lost" errors larger sizes.
i'm seeing end() method of compressor class never called. wonder if might bug in spark lower level library (i'm using spark 1.5.1).
has been down path?
thanks.
Comments
Post a Comment