We using flume ingestig data into HDFS. Flume sink is configured fileType=CompressedStream and codeC=snappy If on any reason flume agent dies or namenode (HA) restarts, flume current sink file will left open - .tmp extension. testfile.snappy.tmp for example.
Is there any way to decompress such file and get data back human readable form? Or is there any tool to fix such files? We can use any other compression too, if there is a way to fix such files. Indrek