How JanusGraph gremlin knows that it should use Cassandra and not Hadoop?

Question

I am trying to use Janusgraph with Cassandra following the guide https://www.bluepiit.com/blog/janusgraph-with-cassandra/ . But I am receiving error while starting gremlin:

C:\Homes\janusgraph-0.2.3-hadoop2\bin>gremlin
HADOOP_HOME is not set.
Download http://public-repo-1.hortonworks.com/hdp-win-alpha/winutils.exe
Place it under C:\Homes\janusgraph-0.2.3-hadoop2\bin\winutils.exe
Press any key to continue . . .

The tutorial (and also the Janusgraph documentation https://docs.janusgraph.org/latest/cassandra.html does not mention that I should set HADOOP_HOME and event if I should, what is the sense to configure Hadoop home if I am willing to use Cassandra? Maybe I should fake the gramlin and set HADOOP_HOME to Cassandra installation? Besides, Janusgraph has configuration files for each of the backends but I can not find the global janusgraph single configuration file in which I could indicate what backend to use?

Aaron Aaron · Accepted Answer · 2019-06-04T13:47:12

Maybe I should fake the gremlin and set HADOOP_HOME to Cassandra installation?

Inside the gremlin.bat I can see the check that you're failing on.

:: Hadoop winutils.exe needs to be available because hadoop-gremlin is installed and active by default
IF NOT DEFINED HADOOP_HOME (
    SET JANUSGRAPH_WINUTILS=%JANUSGRAPH_HOME%\bin\winutils.exe
    IF EXIST !JANUSGRAPH_WINUTILS! (
        SET HADOOP_HOME=%JANUSGRAPH_HOME%
    ) ELSE (
        ECHO HADOOP_HOME is not set.
        ECHO Download http://public-repo-1.hortonworks.com/hdp-win-alpha/winutils.exe
        ECHO Place it under !JANUSGRAPH_WINUTILS!
        PAUSE
        GOTO :eof
    )
)

If you insist on running JanusGraph on Windows, you'll need to follow the line about downloading winutils.exe from hortonworks.com, and copying it to C:\Homes\janusgraph-0.2.3-hadoop2\bin\.

As to getting JanusGraph to use Cassandra, that's something you need to specify in the conf/gremlin/gremlin-server.yaml file.

Specifically, I have set:

channelizer: org.apache.tinkerpop.gremlin.server.channel.WsAndHttpChannelizer
graphs: {
  graph: conf/gremlin-server/janusgraph-cql-server.properties
}

Inside the janusgraph-cql-server.properties file is where you specify your the connection info for your Cassandra cluster.

Then, I would run bin/gremlin-server.bat, instead of gremlin.bat.

Here are some other observations:

Use the latest version of JanusGraph, which I'm pretty sure is 0.3.1.
Connect with CQL instead of Thrift, if you can. The next major version of Cassandra will not even include Thrift, so don't grow attached to it.
Build JanusGraph and Cassandra on Linux. You are setting yourself up for travel on a long road of suffering by using Windows for this.

Hope this helps!

How JanusGraph gremlin knows that it should use Cassandra and not Hadoop?

1 Answers