Description

This processor is used to create a Hadoop Sequence File, which essentially is a file of key/value pairs. The key will be a file name and the value will be the flow file content. The processor will take either a merged (a.k.a. packaged) flow file or a singular flow file. Historically, this processor handled the merging by type and size or time prior to creating a SequenceFile output; it no longer does this. If creating a SequenceFile that contains multiple files of the same type is desired, precede this processor with a RouteOnAttribute processor to segregate files of the same type and follow that with a MergeContent processor to bundle up files. If the type of files is not important, just use the MergeContent processor. When using the MergeContent processor, the following Merge Formats are supported by this processor:

The created SequenceFile is named the same as the incoming FlowFile with the suffix '.sf'. For incoming FlowFiles that are bundled, the keys in the SequenceFile are the individual file names, the values are the contents of each file.

NOTE: The value portion of a key/value pair is loaded into memory. While there is a max size limit of 2GB, this could cause memory issues if there are too many concurrent tasks and the flow file sizes are large.

Using Compression

The value of the Compression codec property determines the compression library the processor uses to compress content. Third party libraries are used for compression. These third party libraries can be Java libraries or native libraries. In case of native libraries, the path of the parent folder needs to be in an environment variable called LD_LIBRARY_PATH so that NiFi can find the libraries.

Example: using Snappy compression with native library on CentOS

  1. Snappy compression needs to be installed on the server running NiFi:
    sudo yum install snappy
  2. Suppose that the server running NiFi has the native compression libraries in /opt/lib/hadoop/lib/native . (Native libraries have file extensions like .so, .dll, .lib, etc. depending on the platform.)
    We need to make sure that the files can be executed by the NiFi process' user. For this purpose we can make a copy of these files to e.g. /opt/nativelibs and change their owner. If NiFi is executed by nifi user in the nifi group, then:
    chown nifi:nifi /opt/nativelibs
    chown nifi:nifi /opt/nativelibs/*
  3. The LD_LIBRARY_PATH needs to be set to contain the path to the folder /opt/nativelibs.
  4. NiFi needs to be restarted.
  5. Compression codec property can be set to SNAPPY and a Compression type can be selected.
  6. The processor can be started.