0

I've a graph with different types of nodes & relationships. Each type of node has 3-4 properties. For testing purpose on HDFS, I'm using GraphSON file to store this graph. Now I want to analyse this graph using Giraph. I've explore Giraph's IO classes & also found that Gremlin can directly load GraphSON. So could you please explain me how to load the graph into Giraph using TinkerPop stack?

Parth
  • 729
  • 8
  • 23

1 Answers1

2

See the Giraph sample in the docs, it does almost exactly what you're looking for. Instead of hadoop-gryo.properties use hadoop-graphson.properties (and of course adjust the input location setting, gremlin.hadoop.inputLocation, in the configuration file).

Daniel Kuppitz
  • 10,846
  • 1
  • 25
  • 34
  • Following the sample, I tried to install `hadoop-gremlin` via **gremlin-console** & got "Error grabbing Grapes -- [unresolved dependency: org.apache.giraph#giraph-core;1.0.0: not found]" – Parth Sep 30 '15 at 16:21
  • Did you adjust your Grapes configuration as described in http://tinkerpop.incubator.apache.org/docs/3.0.1-incubating/#gremlin-applications? – Daniel Kuppitz Sep 30 '15 at 16:29
  • Thanks a lot, I adjusted the config & then I got `==>Loaded: [org.apache.tinkerpop, hadoop-gremlin, 3.0.1-incubating] - restart the console to use [tinkerpop.hadoop]` but after restart, it throws `tinkerpop.hadoop could not be found`. I have got both _hadoop_ & _giraph_ running, so could you please help me connect _hadoop_ and _giraph_ to _gremlin_? – Parth Oct 01 '15 at 12:44
  • Not sure when exactly you get this error. Can you provide a full Gist of your console session? – Daniel Kuppitz Oct 01 '15 at 18:10
  • Hm, that's weird. Maybe something was messed up in one of the previously failed attempts? Please try to remove the `ext/hadoop-gremlin` directory and reinstall the plugin. – Daniel Kuppitz Oct 05 '15 at 12:32
  • Thanks a lot for your patience, but I tried it several times. I even tried to use the plugin without restarting gremlin-console & this was the stack trace: http://hastebin.com/qemaroxotu.avrasm. Do I have to setup hadoop or giraph, because now I'm just using gremlin-console & trying to install hadoop-gremlin plugin? – Parth Oct 06 '15 at 07:45
  • I always have a pseudo-cluster set up locally, but the plugin installation should work without it (though certain things, like the hdfs commands, will not work). – Daniel Kuppitz Oct 06 '15 at 14:01
  • One more thing to consider - did you download the binaries or did you build it from source? If you've done the latter, be sure to start the console from gremlin-console target directory: `./gremlin-console/target/apache-gremlin-console-*-standalone` – Daniel Kuppitz Oct 06 '15 at 18:57