Making JanusGraph work in both OLTP and OLAP in JanusGraph Server with HBase

[John Helmsen]

Due to the considerable efforts of HadoopMarc and others on the board, as well as a great deal of blood sweat and tears from our team here, we have a working implementation of JanusGraph 0.1.1 running on HDP 2.6.4.

This implementation has the following properties:

1) Runs both JanusGraph and SparkGraphComputer.

2) Uses HBase 1.1.2 as its backend.

3) Is addressable from Python.

The general description of how we approached it is listed in: http://yaaics.blogspot.nl/2017/07/configuring-janusgraph-for-spark-yarn.html

The catch, however, is that now we'd like to have this system run as a part of a general program.

To do this, it would be helpful to be able to address all the capabilities as part of a Python program, which requires:

- Running everything through the Gremlin Server, since that is the way that python programs interact with Tinkerpop.

- Having the server somehow have two different methods of addressing the same graph, hopefully through different variables on the server.

- This is because the SparkGraphComputer is not addressable from the JanusGraph instance.  You need to use a HadoopGraph instance to run it.

(If the above is incorrectly assumed, please correct me.)

So, how do we run Spark operations through the server?

[HadoopMarc]

Hi John,

I have never been there, but what did you try? Does anything goes wrong when configuring a HadoopGraph in gremlin server? Any stacktraces?

You may not be aware that you can simply configure multiple graphs in gremlin server. I show two fragments of the .yaml and script files below as an example. From gremlin console each graph is accessible with differently defined Graph and TraversalSource variables.

channelizer: org.apache.tinkerpop.gremlin.server.channel.WebSocketChannelizer
graphs: {
  graph: conf/gremlin-server/titan-berkeleyje-server.properties,
  graph_tst: conf/gremlin-server/tst.properties}
plugins:
  - tinkerpop.tinkergraph
  - aurelius.titan
scriptEngines: {
  gremlin-groovy: {
    imports: [java.lang.Math],
    staticImports: [java.lang.Math.PI],
    scripts: [scripts/empty-sample.groovy],
    config: {
      compilerCustomizerProviders: {
              "org.apache.tinkerpop.gremlin.groovy.jsr223.customizer.ThreadInterruptCustomizerProvider":[],
              "org.apache.tinkerpop.gremlin.groovy.jsr223.customizer.TimedInterruptCustomizerProvider":[10000],
              "org.apache.tinkerpop.gremlin.groovy.jsr223.customizer.CompileStaticCustomizerProvider":["org.apache.tinkerpop.gremlin.groovy.jsr223.customizer.SimpleSandboxExtension"]}}}}

// define the default TraversalSource to bind queries to.
g = graph.traversal()
g_tst = graph_tst.traversal()

HTH,    Marc

Op dinsdag 27 februari 2018 21:08:56 UTC+1 schreef John Helmsen:

[John Helmsen]

Thanks for the response Marc,

What I am seeing so far, is the following.  Here is my yaml file:

host: localhost

port: 8182

scriptEvaluationTimeout: 30000

channelizer: org.apache.tinkerpop.gremlin.server.channel.WebSocketChannelizer

graphs: {

  graph: conf/gremlin-server/janusgraph-stuff-server.properties,

  hgraph: conf/hadoop-graph/hadoop-gryo-stuff.properties}

plugins:

  - janusgraph.imports

scriptEngines: {

  gremlin-groovy: {

    imports: [java.lang.Math],

    staticImports: [java.lang.Math.PI],

    scripts: [scripts/stuff.groovy]},

  gremlin-python: {}}

serializers:

  - { className: org.apache.tinkerpop.gremlin.driver.ser.GryoMessageSerializerV1d0, config: { ioRegistries: [org.janusgraph.graphdb.tinkerpop.JanusGraphIoRegistry] }}

  - { className: org.apache.tinkerpop.gremlin.driver.ser.GryoLiteMessageSerializerV1d0, config: {ioRegistries: [org.janusgraph.graphdb.tinkerpop.JanusGraphIoRegistry] }}

  - { className: org.apache.tinkerpop.gremlin.driver.ser.GryoMessageSerializerV1d0, config: { serializeResultToString: true }}

  - { className: org.apache.tinkerpop.gremlin.driver.ser.GraphSONMessageSerializerGremlinV1d0, config: { ioRegistries: [org.janusgraph.graphdb.tinkerpop.JanusGraphIoRegistry] }}

  - { className: org.apache.tinkerpop.gremlin.driver.ser.GraphSONMessageSerializerGremlinV2d0, config: { ioRegistries: [org.janusgraph.graphdb.tinkerpop.JanusGraphIoRegistry] }}

  - { className: org.apache.tinkerpop.gremlin.driver.ser.GraphSONMessageSerializerV1d0, config: { ioRegistries: [org.janusgraph.graphdb.tinkerpop.JanusGraphIoRegistry] }}

processors:

  - { className: org.apache.tinkerpop.gremlin.server.op.session.SessionOpProcessor, config: { sessionTimeout: 28800000 }}

  - { className: org.apache.tinkerpop.gremlin.server.op.traversal.TraversalOpProcessor, config: { cacheExpirationTime: 600000, cacheMaxSize: 1000 }}

metrics: {

  consoleReporter: {enabled: true, interval: 180000},

  csvReporter: {enabled: true, interval: 180000, fileName: /tmp/gremlin-server-metrics.csv},

  jmxReporter: {enabled: true},

  slf4jReporter: {enabled: true, interval: 180000},

  gangliaReporter: {enabled: false, interval: 180000, addressingMode: MULTICAST},

  graphiteReporter: {enabled: false, interval: 180000}}

maxInitialLineLength: 4096

maxHeaderSize: 8192

maxChunkSize: 8192

maxContentLength: 65536

maxAccumulationBufferComponents: 1024

resultIterationBatchSize: 64

writeBufferLowWaterMark: 32768

writeBufferHighWaterMark: 65536

ssl: {

  enabled: false}

The stuff.groovy script is the following:

// an init script that returns a Map allows explicit setting of global bindings.

def globals = [:]

// defines a sample LifeCycleHook that prints some output to the Gremlin Server console.

// note that the name of the key in the "global" map is unimportant.

globals << [hook : [

        onStartUp: { ctx ->

            ctx.logger.info("Executed once at startup of Gremlin Server.")

        },

        onShutDown: { ctx ->

            ctx.logger.info("Executed once at shutdown of Gremlin Server.")

        }

] as LifeCycleHook]

// define the default TraversalSource to bind queries to - this one will be named "g".

globals << [g : graph.traversal()]

globals << [hg : hgraph.traversal()]

- When looked at through the remote system, both g and hg return as expected.  However, when we try to query with hg:

gremlin> :> hg.V().count()

IllegalStateException

Type ':help' or ':h' for help.

Display stack trace? [yN]

 

and if I try to use the graph computer:

Enter code here...gremlin> :> hg.withComputer(SparkGraphComputer).V().count()

No such property: SparkGraphComputer for class: Script9

Type ':help' or ':h' for help.

Display stack trace? [yN]

So I assume some sort of plugins are not active on the server, and I am going to try to activate those.  If you have any other ideas, please let me know.

[John Helmsen]

Okay, so I added tinkerpop.hadoop and tinkerpop.spark to the plugins list and the following happened:

Enter code here...gremlin> :> hg.V().count()

org.apache.tinkerpop.gremlin.groovy.plugin.RemoteException

Type ':help' or ':h' for help.

Display stack trace? [yN]N

gremlin> :> hg.V().count()

TraversalInterruptedException

Type ':help' or ':h' for help.

Display stack trace? [yN]N

gremlin> :> hg.V().count()

==>12

gremlin> :> hg.V().count()

==>12

So it failed the first two times, but since then it works fine.  It seemed to quit quickly earlier, so perhaps it learned to be patient?

Anyhow, it works for:

g.V().count()

hg.V().count() //Long Yarn Run

g.V().properties()

hg.V().properties() //Also a long yarn run

[John Helmsen]

Okay, so I accessed the server through GremlinPython (which was kind of the whole point of the server) and the following happens:

g.V().count().toList() => [12L]

hg.V().count().toList() => Lots of Yarn, but then [12L]

g.V().properties().toList() => Graph of the Gods Properties

hg.V().properties().toList() => Throws a stack trace for some odd reason in the toList() routine

g.V().valueMap().toList() => Graph of the Gods Properties in a slightly different form

hg.V().valueMap().toList() => Graph of the Gods Properties in the same form as g.V().valueMap().toList()

g.E().<function>.toList() works the same as hg.E().<function>.toList(), including function==properties().

Don't know why only vertex properties are giving trouble.

So clearly, this works, but there are some operations that seem to have a little trouble.  So far, I've found work arounds, so we'll keep pushing forward.

On Wednesday, February 28, 2018 at 2:34:25 AM UTC-5, HadoopMarc wrote: