Another question is how to install graphframes permanently when the spark nodes can not connect to the internet.
------------------ Original ------------------ From: Denny Lee <denny.g....@gmail.com> Date: Mon,Feb 19,2018 10:23 AM To: xiaobo <guxiaobo1...@qq.com> Cc: user@spark.apache.org <user@spark.apache.org> Subject: Re: Does Pyspark Support Graphx? Note the --packages option works for both PySpark and Spark (Scala). For the SparkLauncher class, you should be able to include packages ala: spark.addSparkArg("--packages", "graphframes:0.5.0-spark2.0-s_2.11") On Sun, Feb 18, 2018 at 3:30 PM xiaobo <guxiaobo1...@qq.com> wrote: Hi Denny, The pyspark script uses the --packages option to load graphframe library, what about the SparkLauncher class? ------------------ Original ------------------ From: Denny Lee <denny.g....@gmail.com> Date: Sun,Feb 18,2018 11:07 AM To: 94035420 <guxiaobo1...@qq.com> Cc: user@spark.apache.org <user@spark.apache.org> Subject: Re: Does Pyspark Support Graphx? That??s correct - you can use GraphFrames though as it does support PySpark. On Sat, Feb 17, 2018 at 17:36 94035420 <guxiaobo1...@qq.com> wrote: I can not find anything for graphx module in the python API document, does it mean it is not supported yet?