I was able to set TTL for vertices and edges, but even after the time has passed, the vertex did not got removed from the Db. My storage backend is Cassandra. So do we have to configure anything for Cassandra to activate TTL through janusgraph.

I used the following code :-

graph = JanusGraphFactory.open('conf/janusgraph-cassandra-solr.properties)
mgmt = graph.openManagement()
tweet = mgmt.makeVertexLabel('tweet').setStatic().make()
mgmt.setTTL(tweet, Duration.ofMinutes(2))

I want to know ,one person in the janusGraph ,his relational depth，use gremlin

Hi,

I am exploring Janusgraph bulk load via SparkGraphComputer, janusgraph has been setup as plugin to tinkerpop server and console, with HBase as underlying storage and Elasticsearch as external index store.

I am running this setup on MapR cluster and had to recompile Janusgraph to resolve guava specific conflicts (shaded guava with relocation).

Next I am trying out the example BulkLoaderVertexProgram code provided in Chapter 33, It works fine till I have composite and vertex centric indexes in my schema, but as soon as I define mixed indexes and execute same code I end up with following exception in my Spark Job in stage 2 of job 1 -

java.lang.NoClassDefFoundError: Could not initialize class org.janusgraph.diskstorage.es.ElasticSearchIndex

        at java.lang.Class.forName0(Native Method)

        at java.lang.Class.forName(Class.java:264)

        at org.janusgraph.util.system.ConfigurationUtil.instantiate(ConfigurationUtil.java:56)

        at org.janusgraph.diskstorage.Backend.getImplementationClass(Backend.java:477)

        at org.janusgraph.diskstorage.Backend.getIndexes(Backend.java:464)

        at org.janusgraph.diskstorage.Backend.<init>(Backend.java:149)

        at org.janusgraph.graphdb.configuration.GraphDatabaseConfiguration.getBackend(GraphDatabaseConfiguration.java:1850)

        at org.janusgraph.graphdb.database.StandardJanusGraph.<init>(StandardJanusGraph.java:134)

I have verified that all janusgraph specific jars are in spark executor classpath and mixed indexes work fine with GraphOfGod example.

First I want to understand is it right path to use BulkLoaderVertexProgram be used to add mixed indexes? or should I upload the data and build indexes thereafter?

let me know if any additional info is required to dig deeper.

~mbaxi

On behalf of the JanusGraph Technical Steering Committee (TSC), I'm pleased to welcome 3 new committers on the project! Here they are in alphabetical order by last name.

Robert Dale: Robert has been a solid contributor, and his contributions are across the board -- triaging issues, submitting/reviewing pull requests, and answering questions on the Google groups. He's also on the Apache TinkerPop PMC.

Paul Kendall and Samant Maharaj: Paul and Samant contributed the CQL storage adapter. This is a pretty big achievement and helps steer JanusGraph towards future compatibility with Cassandra 4.0. They are continuing work on cleaning up the Cassandra source code tree that will help make testing it easier and better.

Congratulations to all!

Robert, Paul and Samant - Thanks for the great work you've put into
JanusGraph and welcome aboard!

On Tuesday, August 22, 2017 at 6:32:26 AM UTC-5, Jason Plurad wrote:

On behalf of the JanusGraph Technical Steering Committee (TSC), I'm
pleased to welcome 3 new committers on the project! Here they are in
alphabetical order by last name.

Robert Dale: Robert has been a solid contributor, and his contributions
are across the board -- triaging issues, submitting/reviewing pull requests,
and answering questions on the Google groups. He's also on the Apache
TinkerPop PMC.

Paul Kendall and Samant Maharaj: Paul and Samant contributed the CQL
storage adapter. This is a pretty big achievement and helps steer JanusGraph
towards future compatibility with Cassandra 4.0. They are continuing work on
cleaning up the Cassandra source code tree that will help make testing it
easier and better.

Congratulations to all!

--
You received this message because you are subscribed to the Google Groups
"JanusGraph developers" group.

To unsubscribe from this group and stop receiving emails from it, send an
email to janusgr...@....
For more options, visit https://groups.google.com/d/optout.

Hey - Joseph,Did your test successed?Can you share your experience for me ? Thx

在 2017年8月15日星期二 UTC+8上午6:17:12，Joseph Obernberger写道：

Marc - thank you for this.  I'm going to try getting the latest version of JanusGraph, and compiling it with our specific version of Cloudera CDH, then run some tests.  Will report back.

-Joe

On 8/13/2017 4:07 PM, HadoopMarc wrote:

Hi Joe,

To shed some more light on the running figures you presented, I ran some tests on my own cluster:

1. I loaded the default janusgraph-hbase table with the following simple script from the console:

graph=JanusGraphFactory.open("conf/janusgraph-hbase.properties")
g = graph.traversal()
m = 1200L
n = 10000L
(0L..<m).each{
        (0L..<n).each{
                v1 = g.addV().id().next()
                v2 = g.addV().id().next()
                g.V(v1).addE('link1').to(g.V(v2)).next()
                g.V(v1).addE('link2').to(g.V(v2)).next()
        }
        g.tx().commit()
}

This scipt runs about 20(?) minutes and results in 24M vertices and edges committed to the graph.

2. I did an OLTP g.V().count() on this graph from the console: 11 minutes first time, 10 minutes second time

3. I ran OLAP jobs on this graph using janusgraph-hhbase in two ways:
    a) with g = graph.traversal().withComputer(SparkGraphComputer)  
    b) with g = graph.traversal().withComputer(new Computer().graphComputer(SparkGraphComputer).workers(10))

the properties file was as in the recipe, with the exception of:
   spark.executor.memory=4096m       # smaller values might work, but the 512m from the recipe is definitely too small
   spark.executor.instances=4
   #spark.executor.cores not set, so default value 1

This resulted in the following running times:
   a) stage 0,1,2 => 12min, 12min, 3s => 24min total
   b) stage 0,1,2 => 18min, 1min, 86ms => 19 min total

Discussion:

• HBase is not an easy source for OLAP: HBase wants large regions for efficiency (configurable, but typically 2-20GB), while mapreduce inputformats (like janusgraph's HBaseInputFormat) take regions as inputsplits by default. This means that only a few executors will read from HBase unless the HBaseInputFormat is extended to split a region's keyspace into multiple inputsplits. This mismatch between the numbers of regions and spark executors is a potential JanusGraph issue. Examples exist to improve on this, e.g. org.apache.hadoop.hbase.mapreduce.RowCounter


• For spark stages after stage 0 (reading from HBase), increasing the number of spark tasks with the "workers()" setting helps optimizing the parallelization. This means that for larger traversals than just a vertex count, the parallelization with spark will really pay off.
  
  
• I did not try to repeat your settings with a large number of cores. Various sources discourage the use of spark.executor.cores values larger than 5, e.g. https://blog.cloudera.com/blog/2015/03/how-to-tune-your-apache-spark-jobs-part-2/, https://stackoverflow.com/questions/37871194/how-to-tune-spark-executor-number-cores-and-executor-memory
  

Hopefully, these tests provide you and other readers with some additional perspectives on the configuration of janusgraph-hbase.

Cheers,    Marc

Op donderdag 10 augustus 2017 15:40:21 UTC+2 schreef Joseph Obernberger:

Thank you Marc.

I did not set spark.executor.instances, but I do have spark.cores.max set to 64 and within YARN, it is configured to allow has much RAM/cores for our 5 server cluster.  When I run a job on a table that has 61 regions, I see that 43 tasks are started and running on all 5 nodes in the Spark UI (and running top on each of the servers).  If I lower the amount of RAM (heap) that each tasks has (currently set to 10G), they fail with OutOfMemory exceptions.  It still hits one HBase node very hard and cycles through them.  While that may be a reason for a performance issue, it doesn't explain the massive number of calls that HBase receives for a count job, and why using SparkGraphComputer takes so much more time.

Running with your command below appears to not alter the behavior.  I did run a job last night with DEBUG turned on, but it produced too much logging filling up the log directory on 3 of the 5 nodes before stopping. 
Thanks again Marc!

-Joe

On 8/10/2017 7:33 AM, HadoopMarc wrote:

Hi Joe,

Another thing to try (only tested on Tinkerpop, not on JanusGraph): create the traversalsource as follows:

g = graph.traversal().withComputer(new Computer().graphComputer(SparkGraphComputer).workers(100))

With HadoopGraph this helps hdfs files with very large or no partitions to be split across tasks; I did not check the effect yet for HBaseInputFormat in JanusGraph. And did you add spark.executor.instances=10 (or some suitable number) to your config? And did you check in the RM ui or Spark history server whether these executors were really allocated and started?

More later,

Marc

Op donderdag 10 augustus 2017 00:13:09 UTC+2 schreef Joseph Obernberger:

Marc - thank you.  I've updated the classpath and removed nearly all of the CDH jars; had to keep chimera and some of the HBase libs in there.  Apart from those and all the jars in lib.zip, it is working as it did before.  The reason I turned DEBUG off was because it was producing 100+GBytes of logs.  Nearly all of which are things like:

18:04:29 DEBUG org.janusgraph.diskstorage.hbase.HBaseKeyColumnValueStore - Generated HBase Filter ColumnRangeFilter [\x10\xC0, \x10\xC1)
18:04:29 DEBUG org.janusgraph.graphdb.transaction.StandardJanusGraphTx - Guava vertex cache size: requested=20000 effective=20000 (min=100)
18:04:29 DEBUG org.janusgraph.graphdb.transaction.vertexcache.GuavaVertexCache - Created dirty vertex map with initial size 32
18:04:29 DEBUG org.janusgraph.graphdb.transaction.vertexcache.GuavaVertexCache - Created vertex cache with max size 20000
18:04:29 DEBUG org.janusgraph.diskstorage.hbase.HBaseKeyColumnValueStore - Generated HBase Filter ColumnRangeFilter [\x10\xC2, \x10\xC3)
18:04:29 DEBUG org.janusgraph.graphdb.transaction.StandardJanusGraphTx - Guava vertex cache size: requested=20000 effective=20000 (min=100)
18:04:29 DEBUG org.janusgraph.graphdb.transaction.vertexcache.GuavaVertexCache - Created dirty vertex map with initial size 32
18:04:29 DEBUG org.janusgraph.graphdb.transaction.vertexcache.GuavaVertexCache - Created vertex cache with max size 20000

Do those mean anything to you?  I've turned it back on for running with smaller graph sizes, but so far I don't see anything helpful there apart from an exception about not setting HADOOP_HOME.
Here are the spark properties; notice the nice and small extraClassPath!  :)

Name	Value
gremlin.graph	org.apache.tinkerpop.gremlin.hadoop.structure.HadoopGraph
gremlin.hadoop.deriveMemory	false
gremlin.hadoop.graphReader	org.janusgraph.hadoop.formats.hbase.HBaseInputFormat
gremlin.hadoop.graphWriter	org.apache.tinkerpop.gremlin.hadoop.structure.io.gryo.GryoOutputFormat
gremlin.hadoop.graphWriter.hasEdges	false
gremlin.hadoop.inputLocation	none
gremlin.hadoop.jarsInDistributedCache	true
gremlin.hadoop.memoryOutputFormat	org.apache.tinkerpop.gremlin.hadoop.structure.io.gryo.GryoOutputFormat
gremlin.hadoop.outputLocation	output
janusgraphmr.ioformat.conf.storage.backend	hbase
janusgraphmr.ioformat.conf.storage.hbase.region-count	5
janusgraphmr.ioformat.conf.storage.hbase.regions-per-server	5
janusgraphmr.ioformat.conf.storage.hbase.short-cf-names	false
janusgraphmr.ioformat.conf.storage.hbase.table	TEST0.2.0
janusgraphmr.ioformat.conf.storage.hostname	10.22.5.65:2181
log4j.appender.STDOUT	org.apache.log4j.ConsoleAppender
log4j.logger.deng	WARNING
log4j.rootLogger	STDOUT
org.slf4j.simpleLogger.defaultLogLevel	warn
spark.akka.frameSize	1024
spark.app.id	application_1502118729859_0041
spark.app.name	Apache TinkerPop's Spark-Gremlin
spark.authenticate	false
spark.cores.max	64
spark.driver.appUIAddress	http://10.22.5.61:4040
spark.driver.extraJavaOptons	-XX:ReservedCodeCacheSize=100M -XX:MaxMetaspaceSize=256m -XX:CompressedClassSpaceSize=256m
spark.driver.extraLibraryPath	/opt/cloudera/parcels/CDH-5.10.0-1.cdh5.10.0.p0.41/lib/hadoop/lib/native
spark.driver.host	10.22.5.61
spark.driver.port	38529
spark.dynamicAllocation.enabled	true
spark.dynamicAllocation.executorIdleTimeout	60
spark.dynamicAllocation.minExecutors	0
spark.dynamicAllocation.schedulerBacklogTimeout	1
spark.eventLog.dir	hdfs://host001:8020/user/spark/applicationHistory
spark.eventLog.enabled	true
spark.executor.extraClassPath	/opt/cloudera/parcels/CDH/jars/janusgraph-hbase-0.2.0-SNAPSHOT.jar:./lib.zip/*:/opt/cloudera/parcels/CDH/lib/hbase/bin/../lib/*:/etc/hbase/conf:
spark.executor.extraJavaOptions	-XX:ReservedCodeCacheSize=100M -XX:MaxMetaspaceSize=256m -XX:CompressedClassSpaceSize=256m -Dlogback.configurationFile=logback.xml
spark.executor.extraLibraryPath	/opt/cloudera/parcels/CDH-5.10.0-1.cdh5.10.0.p0.41/lib/hadoop/lib/native
spark.executor.heartbeatInterval	100000
spark.executor.id	driver
spark.executor.memory	10240m
spark.externalBlockStore.folderName	spark-27dac3f3-dfbc-4f32-b52d-ececdbcae0db
spark.kyroserializer.buffer.max	1600m
spark.master	yarn-client
spark.network.timeout	90000
spark.org.apache.hadoop.yarn.server.webproxy.amfilter.AmIpFilter.param.PROXY_HOSTS	host005
spark.org.apache.hadoop.yarn.server.webproxy.amfilter.AmIpFilter.param.PROXY_URI_BASES	http://host005:8088/proxy/application_1502118729859_0041
spark.scheduler.mode	FIFO
spark.serializer	org.apache.spark.serializer.KryoSerializer
spark.shuffle.service.enabled	true
spark.shuffle.service.port	7337
spark.ui.filters	org.apache.hadoop.yarn.server.webproxy.amfilter.AmIpFilter
spark.ui.killEnabled	true
spark.yarn.am.extraLibraryPath	/opt/cloudera/parcels/CDH-5.10.0-1.cdh5.10.0.p0.41/lib/hadoop/lib/native
spark.yarn.appMasterEnv.CLASSPATH	/etc/haddop/conf:/etc/hbase/conf:./lib.zip/*
spark.yarn.config.gatewayPath	/opt/cloudera/parcels
spark.yarn.config.replacementPath	{{HADOOP_COMMON_HOME}}/../../..
spark.yarn.dist.archives	/home/graph/janusgraph-0.2.0-SNAPSHOT-hadoop2.JOE/lib.zip
spark.yarn.dist.files	/home/graph/janusgraph-0.2.0-SNAPSHOT-hadoop2.JOE/conf/logback.xml
spark.yarn.dist.jars	/opt/cloudera/parcels/CDH/jars/janusgraph-hbase-0.2.0-SNAPSHOT.jar
spark.yarn.historyServer.address	http://host001:18088
zookeeper.znode.parent	/hbase

-Joe

On 8/9/2017 3:33 PM, HadoopMarc wrote:

Hi Gari and Joe,

Glad to see you testing the recipes for MapR and Cloudera respectively!  I am sure that you realized by now that getting this to work is like walking through a minefield. If you deviate from the known path, the odds for getting through are dim, and no one wants to be in your vicinity. So, if you see a need to deviate (which there may be for the hadoop distributions you use), you will need your mine sweeper, that is, put the logging level to DEBUG for relevant java packages.

This is where you deviated:

• for Gari: you put all kinds of MapR lib folders on the applications master's classpath (other classpath configs are not visible from your post)
  
• for Joe: you put all kinds of Cloudera lib folders on the executors classpath (worst of all the spark-assembly.jar)

Probably, you experience all kinds of mismatches in netty libraries which slows down or even kills all comms between the yarn containers. The philosophy of the recipes really is to only add the minimum number of conf folders and jars to the Tinkerpop/Janusgraph distribution and see from there if any libraries are missing.

At my side, it has become apparent that I should at least add to the recipes:

• proof of work for a medium-sized graph (say 10M vertices and edges)
  
• configs for the number of executors present in the OLAP job (instead of relying on spark default number of 2)
  

So, still some work to do!

Cheers,    Marc

--
You received this message because you are subscribed to the Google Groups "JanusGraph users list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to janusgraph-use...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Virus-free. www.avg.com

--
You received this message because you are subscribed to the Google Groups "JanusGraph users list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to janusgraph-use...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "JanusGraph users list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to janusgra...@....
For more options, visit https://groups.google.com/d/optout.

On behalf of the JanusGraph Technical Steering Committee (TSC), I'm pleased to welcome 3 new committers on the project! Here they are in alphabetical order by last name.

Robert Dale: Robert has been a solid contributor, and his contributions are across the board -- triaging issues, submitting/reviewing pull requests, and answering questions on the Google groups. He's also on the Apache TinkerPop PMC.

Paul Kendall and Samant Maharaj: Paul and Samant contributed the CQL storage adapter. This is a pretty big achievement and helps steer JanusGraph towards future compatibility with Cassandra 4.0. They are continuing work on cleaning up the Cassandra source code tree that will help make testing it easier and better.

Congratulations to all!

Hi Takao,

JanusGraph reads data from distributed backends into hadoop using its HBaseInputFormat and CassandraInputFomat classes (which are descendents of org.apache.hadoop.mapreduce.InputFormat). Therefore, it seems possible to directly access graphs in these backends from spark using sc.newAPIHadoopRDD. AFAIK, this particular use of the inputformats is nowhere documented or demonstrated, though.

My earlier answer effectively came down to storing the graph to hdfs using the OutputRDD class for the gremlin.hadoop.graphWriter property and spark serialization (my earlier suggestion of persisting the graphRDD using PersistedOutputRDD would not work for you because python and gremlin-server would not share the same SparkContext). This may or may not be easier or more efficient than writing your own csv input/output routines (in combination with the BulkDumperVertexProgram to parallelize the writing).

Hope this helps,

Marc



Op vrijdag 18 augustus 2017 04:19:33 UTC+2 schreef Takao Magoori:

Hi Marc,

Thank you!

But I don't understand what you mean, sorry.

I feel SparkGraphComputer is "OLAP by gremlin on top of spark distributed power". But I want "OLAP by spark using janusGraph data".

So, I want to run "spark-submit", create pyspark sparkContext, load JanusGraph data into DataFrame. Then, I can use spark Dataframe, spark ML and python machine-learning packages.

The following pseudo-code is what really I want. (like https://github.com/sbcd90/spark-orientdb)

If there is no such solution, I guess I have to "dump whole graph into csv and read it from pyspark".

--------

spark_session = SparkSession.builder.appName('test').enableHiveSupport().getOrCreate()

df_user = spark_session.read.format(

    'org.apache.janusgraph.some_spark_gremlin_connector',

).options(

    url='url',

    query='g.V().hasLabel("user").has("age", gt(29)).valueMap("user_id", "name" "age")',

).load().dropna().join(

    other=some_df,

)

df_item = spark_session.read.format(

    'org.apache.janusgraph.some_spark_gremlin_connector',

).options(

    url='url',

    query='g.V().hasLabel("user").has("age", gt(29)).out("buy").hasLabel("item").valueMap("item_id", "name")',

).load().dropna()

df_sale = spark_session.read.format(

    'org.apache.janusgraph.some_spark_gremlin_connector',

).options(

    url='url',

    query='g.V().hasLabel("user").has("age", gt(29)).outE("buy").valueMap("timestamp")',

).load().select(

    col('item_id'),

    col('name'),

).dropna()

--------

2017年8月18日金曜日 4時08分02秒 UTC+9 HadoopMarc:

Hi Takao,

Only some directions. If you combine:

http://yaaics.blogspot.nl/              (using CassandraInputFormat in your case)
http://tinkerpop.apache.org/docs/current/reference/#interacting-with-spark

it should be possible to access the PersistedInputRDD alias graphRDD from the Spark object.
Never done this myself, I would be interested to read if this works! Probably you will need to run an OLAP query with SparkGraphComputer anyway (e.g. g.V()) to have the PersistedInputRDD realized (RDD's are not realized until a spark action is run on them.)

Cheers,     Marc


Op donderdag 17 augustus 2017 16:25:42 UTC+2 schreef Takao Magoori:

I have a JanusGraph Server (github master, gremlin 3.2.5) on top of Cassandra storage backend, to store users, items and "WHEN, WHERE, WHO bought WHAT ?" relations.

To get data from and modify data in the graph, I use Python aiogremlin driver-mode (== groovy sessionless eval mode) and it works well for now. Thanks developers !

So now, I have to compute recommendation and forecast item sales.

In order to data-cleaning, data-normalization, recommendation and forecasting, Because of a little big graph, I want to use higher-level pyspark tools (ex. DataFrame, ML) and python machine learning packages (ex, scikit-learn). But I can not find the way to load graph data into Spark. What I want is "connector" which can be used by pyspark to load data from JanusGraph, not SparkGraphComputer.

Could someone please how to do it ?

- Additional info

It seems OrientDB has some Spark connectors (though, I don't know these can be used by pyspark). But I want JanusGraph's one.

https://github.com/sbcd90/spark-orientdb

https://github.com/metreta/spark-orientdb-connector

I used 2 methods to import it all are failed.

gremlin> graph.io(graphson()).readGraph("import/test.json")
graph.io(IoCore.graphson()).readGraph("import/test.json");
But for the example graphson I can import it.

gremlin> graph.io(graphson()).readGraph("data/tinkerpop-modern.json")
Another issue is about update and delete vertices and edges?

Below is the failed GraphSON file example:
This is the reference :
https://github.com/tinkerpop/blueprints/wiki/GraphSON-Reader-and-Writer-Library
{
    "graph": {
        "mode":"NORMAL",
        "vertices": [
            {
                "name": "lop",
                "lang": "java",
                "_id": "3",
                "_type": "vertex"
            },
            {
                "name": "vadas",
                "age": 27,
                "_id": "2",
                "_type": "vertex"
            },
            {
                "name": "marko",
                "age": 29,
                "_id": "1",
                "_type": "vertex"
            },
            {
                "name": "peter",
                "age": 35,
                "_id": "6",
                "_type": "vertex"
            },
            {
                "name": "ripple",
                "lang": "java",
                "_id": "5",
                "_type": "vertex"
            },
            {
                "name": "josh",
                "age": 32,
                "_id": "4",
                "_type": "vertex"
            }
        ],
        "edges": [
            {
                "weight": 1,
                "_id": "10",
                "_type": "edge",
                "_outV": "4",
                "_inV": "5",
                "_label": "created"
            },
            {
                "weight": 0.5,
                "_id": "7",
                "_type": "edge",
                "_outV": "1",
                "_inV": "2",
                "_label": "knows"
            },
            {
                "weight": 0.4000000059604645,
                "_id": "9",
                "_type": "edge",
                "_outV": "1",
                "_inV": "3",
                "_label": "created"
            },
            {
                "weight": 1,
                "_id": "8",
                "_type": "edge",
                "_outV": "1",
                "_inV": "4",
                "_label": "knows"
            },
            {
                "weight": 0.4000000059604645,
                "_id": "11",
                "_type": "edge",
                "_outV": "4",
                "_inV": "3",
                "_label": "created"
            },
            {
                "weight": 0.20000000298023224,
                "_id": "12",
                "_type": "edge",
                "_outV": "6",
                "_inV": "3",
                "_label": "created"
            }
        ]
    }
}

That is an outdated version of TinkerPop and GraphSON as stated on that page.  Use this current reference  http://tinkerpop.apache.org/docs/current/reference/#graphson-reader-writer

Robert Dale

On Mon, Aug 7, 2017 at 1:17 AM, hu junjie <h...@...> wrote:

I used 2 methods to import it all are failed.

gremlin> graph.io(graphson()).readGraph("import/test.json")
graph.io(IoCore.graphson()).readGraph("import/test.json");
But for the example graphson I can import it.

gremlin> graph.io(graphson()).readGraph("data/tinkerpop-modern.json")
Another issue is about update and delete vertices and edges?

Below is the failed GraphSON file example:
This is the reference :
https://github.com/tinkerpop/blueprints/wiki/GraphSON-Reader-and-Writer-Library
{
    "graph": {
        "mode":"NORMAL",
        "vertices": [
            {
                "name": "lop",
                "lang": "java",
                "_id": "3",
                "_type": "vertex"
            },
            {
                "name": "vadas",
                "age": 27,
                "_id": "2",
                "_type": "vertex"
            },
            {
                "name": "marko",
                "age": 29,
                "_id": "1",
                "_type": "vertex"
            },
            {
                "name": "peter",
                "age": 35,
                "_id": "6",
                "_type": "vertex"
            },
            {
                "name": "ripple",
                "lang": "java",
                "_id": "5",
                "_type": "vertex"
            },
            {
                "name": "josh",
                "age": 32,
                "_id": "4",
                "_type": "vertex"
            }
        ],
        "edges": [
            {
                "weight": 1,
                "_id": "10",
                "_type": "edge",
                "_outV": "4",
                "_inV": "5",
                "_label": "created"
            },
            {
                "weight": 0.5,
                "_id": "7",
                "_type": "edge",
                "_outV": "1",
                "_inV": "2",
                "_label": "knows"
            },
            {
                "weight": 0.4000000059604645,
                "_id": "9",
                "_type": "edge",
                "_outV": "1",
                "_inV": "3",
                "_label": "created"
            },
            {
                "weight": 1,
                "_id": "8",
                "_type": "edge",
                "_outV": "1",
                "_inV": "4",
                "_label": "knows"
            },
            {
                "weight": 0.4000000059604645,
                "_id": "11",
                "_type": "edge",
                "_outV": "4",
                "_inV": "3",
                "_label": "created"
            },
            {
                "weight": 0.20000000298023224,
                "_id": "12",
                "_type": "edge",
                "_outV": "6",
                "_inV": "3",
                "_label": "created"
            }
        ]
    }
}

--
You received this message because you are subscribed to the Google Groups "JanusGraph users list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to janusgraph-use...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Marc - thank you for this.  I'm going to try getting the latest version of JanusGraph, and compiling it with our specific version of Cloudera CDH, then run some tests.  Will report back.

-Joe

On 8/13/2017 4:07 PM, HadoopMarc wrote:

Hi Joe,

To shed some more light on the running figures you presented, I ran some tests on my own cluster:

1. I loaded the default janusgraph-hbase table with the following simple script from the console:

graph=JanusGraphFactory.open("conf/janusgraph-hbase.properties")
g = graph.traversal()
m = 1200L
n = 10000L
(0L..<m).each{
        (0L..<n).each{
                v1 = g.addV().id().next()
                v2 = g.addV().id().next()
                g.V(v1).addE('link1').to(g.V(v2)).next()
                g.V(v1).addE('link2').to(g.V(v2)).next()
        }
        g.tx().commit()
}

This scipt runs about 20(?) minutes and results in 24M vertices and edges committed to the graph.

2. I did an OLTP g.V().count() on this graph from the console: 11 minutes first time, 10 minutes second time

3. I ran OLAP jobs on this graph using janusgraph-hhbase in two ways:
    a) with g = graph.traversal().withComputer(SparkGraphComputer)  
    b) with g = graph.traversal().withComputer(new Computer().graphComputer(SparkGraphComputer).workers(10))

the properties file was as in the recipe, with the exception of:
   spark.executor.memory=4096m       # smaller values might work, but the 512m from the recipe is definitely too small
   spark.executor.instances=4
   #spark.executor.cores not set, so default value 1

This resulted in the following running times:
   a) stage 0,1,2 => 12min, 12min, 3s => 24min total
   b) stage 0,1,2 => 18min, 1min, 86ms => 19 min total

Discussion:

• HBase is not an easy source for OLAP: HBase wants large regions for efficiency (configurable, but typically 2-20GB), while mapreduce inputformats (like janusgraph's HBaseInputFormat) take regions as inputsplits by default. This means that only a few executors will read from HBase unless the HBaseInputFormat is extended to split a region's keyspace into multiple inputsplits. This mismatch between the numbers of regions and spark executors is a potential JanusGraph issue. Examples exist to improve on this, e.g. org.apache.hadoop.hbase.mapreduce.RowCounter


• For spark stages after stage 0 (reading from HBase), increasing the number of spark tasks with the "workers()" setting helps optimizing the parallelization. This means that for larger traversals than just a vertex count, the parallelization with spark will really pay off.
  
  
• I did not try to repeat your settings with a large number of cores. Various sources discourage the use of spark.executor.cores values larger than 5, e.g. https://blog.cloudera.com/blog/2015/03/how-to-tune-your-apache-spark-jobs-part-2/, https://stackoverflow.com/questions/37871194/how-to-tune-spark-executor-number-cores-and-executor-memory
  

Hopefully, these tests provide you and other readers with some additional perspectives on the configuration of janusgraph-hbase.

Cheers,    Marc

Op donderdag 10 augustus 2017 15:40:21 UTC+2 schreef Joseph Obernberger:

Thank you Marc.

I did not set spark.executor.instances, but I do have spark.cores.max set to 64 and within YARN, it is configured to allow has much RAM/cores for our 5 server cluster.  When I run a job on a table that has 61 regions, I see that 43 tasks are started and running on all 5 nodes in the Spark UI (and running top on each of the servers).  If I lower the amount of RAM (heap) that each tasks has (currently set to 10G), they fail with OutOfMemory exceptions.  It still hits one HBase node very hard and cycles through them.  While that may be a reason for a performance issue, it doesn't explain the massive number of calls that HBase receives for a count job, and why using SparkGraphComputer takes so much more time.

Running with your command below appears to not alter the behavior.  I did run a job last night with DEBUG turned on, but it produced too much logging filling up the log directory on 3 of the 5 nodes before stopping. 
Thanks again Marc!

-Joe

On 8/10/2017 7:33 AM, HadoopMarc wrote:

Hi Joe,

Another thing to try (only tested on Tinkerpop, not on JanusGraph): create the traversalsource as follows:

g = graph.traversal().withComputer(new Computer().graphComputer(SparkGraphComputer).workers(100))

With HadoopGraph this helps hdfs files with very large or no partitions to be split across tasks; I did not check the effect yet for HBaseInputFormat in JanusGraph. And did you add spark.executor.instances=10 (or some suitable number) to your config? And did you check in the RM ui or Spark history server whether these executors were really allocated and started?

More later,

Marc

Op donderdag 10 augustus 2017 00:13:09 UTC+2 schreef Joseph Obernberger:

Marc - thank you.  I've updated the classpath and removed nearly all of the CDH jars; had to keep chimera and some of the HBase libs in there.  Apart from those and all the jars in lib.zip, it is working as it did before.  The reason I turned DEBUG off was because it was producing 100+GBytes of logs.  Nearly all of which are things like:

18:04:29 DEBUG org.janusgraph.diskstorage.hbase.HBaseKeyColumnValueStore - Generated HBase Filter ColumnRangeFilter [\x10\xC0, \x10\xC1)
18:04:29 DEBUG org.janusgraph.graphdb.transaction.StandardJanusGraphTx - Guava vertex cache size: requested=20000 effective=20000 (min=100)
18:04:29 DEBUG org.janusgraph.graphdb.transaction.vertexcache.GuavaVertexCache - Created dirty vertex map with initial size 32
18:04:29 DEBUG org.janusgraph.graphdb.transaction.vertexcache.GuavaVertexCache - Created vertex cache with max size 20000
18:04:29 DEBUG org.janusgraph.diskstorage.hbase.HBaseKeyColumnValueStore - Generated HBase Filter ColumnRangeFilter [\x10\xC2, \x10\xC3)
18:04:29 DEBUG org.janusgraph.graphdb.transaction.StandardJanusGraphTx - Guava vertex cache size: requested=20000 effective=20000 (min=100)
18:04:29 DEBUG org.janusgraph.graphdb.transaction.vertexcache.GuavaVertexCache - Created dirty vertex map with initial size 32
18:04:29 DEBUG org.janusgraph.graphdb.transaction.vertexcache.GuavaVertexCache - Created vertex cache with max size 20000

Do those mean anything to you?  I've turned it back on for running with smaller graph sizes, but so far I don't see anything helpful there apart from an exception about not setting HADOOP_HOME.
Here are the spark properties; notice the nice and small extraClassPath!  :)

Name	Value
gremlin.graph	org.apache.tinkerpop.gremlin.hadoop.structure.HadoopGraph
gremlin.hadoop.deriveMemory	false
gremlin.hadoop.graphReader	org.janusgraph.hadoop.formats.hbase.HBaseInputFormat
gremlin.hadoop.graphWriter	org.apache.tinkerpop.gremlin.hadoop.structure.io.gryo.GryoOutputFormat
gremlin.hadoop.graphWriter.hasEdges	false
gremlin.hadoop.inputLocation	none
gremlin.hadoop.jarsInDistributedCache	true
gremlin.hadoop.memoryOutputFormat	org.apache.tinkerpop.gremlin.hadoop.structure.io.gryo.GryoOutputFormat
gremlin.hadoop.outputLocation	output
janusgraphmr.ioformat.conf.storage.backend	hbase
janusgraphmr.ioformat.conf.storage.hbase.region-count	5
janusgraphmr.ioformat.conf.storage.hbase.regions-per-server	5
janusgraphmr.ioformat.conf.storage.hbase.short-cf-names	false
janusgraphmr.ioformat.conf.storage.hbase.table	TEST0.2.0
janusgraphmr.ioformat.conf.storage.hostname	10.22.5.65:2181
log4j.appender.STDOUT	org.apache.log4j.ConsoleAppender
log4j.logger.deng	WARNING
log4j.rootLogger	STDOUT
org.slf4j.simpleLogger.defaultLogLevel	warn
spark.akka.frameSize	1024
spark.app.id	application_1502118729859_0041
spark.app.name	Apache TinkerPop's Spark-Gremlin
spark.authenticate	false
spark.cores.max	64
spark.driver.appUIAddress	http://10.22.5.61:4040
spark.driver.extraJavaOptons	-XX:ReservedCodeCacheSize=100M -XX:MaxMetaspaceSize=256m -XX:CompressedClassSpaceSize=256m
spark.driver.extraLibraryPath	/opt/cloudera/parcels/CDH-5.10.0-1.cdh5.10.0.p0.41/lib/hadoop/lib/native
spark.driver.host	10.22.5.61
spark.driver.port	38529
spark.dynamicAllocation.enabled	true
spark.dynamicAllocation.executorIdleTimeout	60
spark.dynamicAllocation.minExecutors	0
spark.dynamicAllocation.schedulerBacklogTimeout	1
spark.eventLog.dir	hdfs://host001:8020/user/spark/applicationHistory
spark.eventLog.enabled	true
spark.executor.extraClassPath	/opt/cloudera/parcels/CDH/jars/janusgraph-hbase-0.2.0-SNAPSHOT.jar:./lib.zip/*:/opt/cloudera/parcels/CDH/lib/hbase/bin/../lib/*:/etc/hbase/conf:
spark.executor.extraJavaOptions	-XX:ReservedCodeCacheSize=100M -XX:MaxMetaspaceSize=256m -XX:CompressedClassSpaceSize=256m -Dlogback.configurationFile=logback.xml
spark.executor.extraLibraryPath	/opt/cloudera/parcels/CDH-5.10.0-1.cdh5.10.0.p0.41/lib/hadoop/lib/native
spark.executor.heartbeatInterval	100000
spark.executor.id	driver
spark.executor.memory	10240m
spark.externalBlockStore.folderName	spark-27dac3f3-dfbc-4f32-b52d-ececdbcae0db
spark.kyroserializer.buffer.max	1600m
spark.master	yarn-client
spark.network.timeout	90000
spark.org.apache.hadoop.yarn.server.webproxy.amfilter.AmIpFilter.param.PROXY_HOSTS	host005
spark.org.apache.hadoop.yarn.server.webproxy.amfilter.AmIpFilter.param.PROXY_URI_BASES	http://host005:8088/proxy/application_1502118729859_0041
spark.scheduler.mode	FIFO
spark.serializer	org.apache.spark.serializer.KryoSerializer
spark.shuffle.service.enabled	true
spark.shuffle.service.port	7337
spark.ui.filters	org.apache.hadoop.yarn.server.webproxy.amfilter.AmIpFilter
spark.ui.killEnabled	true
spark.yarn.am.extraLibraryPath	/opt/cloudera/parcels/CDH-5.10.0-1.cdh5.10.0.p0.41/lib/hadoop/lib/native
spark.yarn.appMasterEnv.CLASSPATH	/etc/haddop/conf:/etc/hbase/conf:./lib.zip/*
spark.yarn.config.gatewayPath	/opt/cloudera/parcels
spark.yarn.config.replacementPath	{{HADOOP_COMMON_HOME}}/../../..
spark.yarn.dist.archives	/home/graph/janusgraph-0.2.0-SNAPSHOT-hadoop2.JOE/lib.zip
spark.yarn.dist.files	/home/graph/janusgraph-0.2.0-SNAPSHOT-hadoop2.JOE/conf/logback.xml
spark.yarn.dist.jars	/opt/cloudera/parcels/CDH/jars/janusgraph-hbase-0.2.0-SNAPSHOT.jar
spark.yarn.historyServer.address	http://host001:18088
zookeeper.znode.parent	/hbase

-Joe

On 8/9/2017 3:33 PM, HadoopMarc wrote:

Hi Gari and Joe,

Glad to see you testing the recipes for MapR and Cloudera respectively!  I am sure that you realized by now that getting this to work is like walking through a minefield. If you deviate from the known path, the odds for getting through are dim, and no one wants to be in your vicinity. So, if you see a need to deviate (which there may be for the hadoop distributions you use), you will need your mine sweeper, that is, put the logging level to DEBUG for relevant java packages.

This is where you deviated:

• for Gari: you put all kinds of MapR lib folders on the applications master's classpath (other classpath configs are not visible from your post)
  
• for Joe: you put all kinds of Cloudera lib folders on the executors classpath (worst of all the spark-assembly.jar)

Probably, you experience all kinds of mismatches in netty libraries which slows down or even kills all comms between the yarn containers. The philosophy of the recipes really is to only add the minimum number of conf folders and jars to the Tinkerpop/Janusgraph distribution and see from there if any libraries are missing.

At my side, it has become apparent that I should at least add to the recipes:

• proof of work for a medium-sized graph (say 10M vertices and edges)
  
• configs for the number of executors present in the OLAP job (instead of relying on spark default number of 2)
  

So, still some work to do!

Cheers,    Marc

--
You received this message because you are subscribed to the Google Groups "JanusGraph users list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to janusgraph-use...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Virus-free. www.avg.com

--
You received this message because you are subscribed to the Google Groups "JanusGraph users list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to janusgraph-use...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

package com.sankuai.kg;


import java.io.File;
import java.util.Iterator;


import org.apache.commons.io.FileUtils;
import org.apache.tinkerpop.gremlin.process.traversal.dsl.graph.GraphTraversal;
import org.apache.tinkerpop.gremlin.process.traversal.dsl.graph.GraphTraversalSource;
import org.apache.tinkerpop.gremlin.structure.Graph;
import org.apache.tinkerpop.gremlin.structure.Transaction;
import org.apache.tinkerpop.gremlin.structure.Vertex;
import org.apache.tinkerpop.gremlin.structure.util.empty.EmptyGraph;


public class Loader {


    public static void main(String[] args) throws Exception {
        Graph graph = EmptyGraph.instance();
        GraphTraversalSource g = graph.traversal().withRemote("remote-graph.properties");
        Iterator<String> lineIt = FileUtils.lineIterator(new File(args[0]));
        while (lineIt.hasNext()) {
            String line = lineIt.next();
            String[] cols = line.split(",");
           
            GraphTraversal<Vertex, Vertex> t1 = g.V().has("poiId", cols[0]);
            GraphTraversal<Vertex, Vertex> t2 = g.V().has("poiId", cols[3]);


            if (!t1.hasNext())
                g.addV().property("poiId", cols[0]).property("name", cols[1]).property("type", cols[2]).next();
            if (!t2.hasNext())
                g.addV().property(String.format("propId", cols[5]), cols[3]).property("name", cols[4])
                        .property("type", cols[5]).next();
            g.V().has("poiId", cols[0]).as("a").V().has("propId", cols[3]).as("b").addE(cols[6])
                    .from("a").to("b").next();
           
        }


        g.close();
    }


}

Maybe search to see if there's a known issue with running Cassandra on a MacBook. You could upgrade Cassandra if that's what you need. I believe JanusGraph is known to work with all current versions.

Robert Dale

On Fri, Aug 18, 2017 at 12:06 PM, 'Ray Scott' via JanusGraph users list <janusgra...@googlegroups.com> wrote:

As soon as I killed the cassandra process, the CPU usage plummeted. So at least I know who the culprit was. I'll just start Gremlin Server directly (configured to us BDB), instead of using JanusGraph.sh. 

On Thursday, 17 August 2017 19:22:51 UTC+1, Robert Dale wrote:

For how long does the cpu remain high after you get the `gremlin>` prompt?

Robert Dale

On Thu, Aug 17, 2017 at 2:20 PM, 'Ray Scott' via JanusGraph users list <janu...@...> wrote:

Hi, 

I'm trying to get JanusGraph running on my laptop (MacBook Air 2 Core Intel i7, 8GB) so that I can develop a small working prototype. 

I've gone for the janusgraph.sh method of starting the server and everything is fine until I open a gremlin shell. Then I see the CPU usage for my terminal rocket up past 350%. Once I close the gremlin shell, the CPU usage remains at the same high level, indefinitely. I've tried launching with the Berkeley DB option and no Elastic Search, but I get the exact same behaviour. 

Is there something I can do to stop this from using most of my CPU, or do I just have to live with it? 

Is there a "lite" recommended setup that someone has had success with in the past? 

Thanks. 

--
You received this message because you are subscribed to the Google Groups "JanusGraph users list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to janusgraph-use...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "JanusGraph users list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to janusgraph-use...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Hi Marc,

Thank you!

But I don't understand what you mean, sorry.

I feel SparkGraphComputer is "OLAP by gremlin on top of spark distributed power". But I want "OLAP by spark using janusGraph data".

So, I want to run "spark-submit", create pyspark sparkContext, load JanusGraph data into DataFrame. Then, I can use spark Dataframe, spark ML and python machine-learning packages.

The following pseudo-code is what really I want. (like https://github.com/sbcd90/spark-orientdb)

If there is no such solution, I guess I have to "dump whole graph into csv and read it from pyspark".

--------

spark_session = SparkSession.builder.appName('test').enableHiveSupport().getOrCreate()

df_user = spark_session.read.format(

    'org.apache.janusgraph.some_spark_gremlin_connector',

).options(

    url='url',

    query='g.V().hasLabel("user").has("age", gt(29)).valueMap("user_id", "name" "age")',

).load().dropna().join(

    other=some_df,

)

df_item = spark_session.read.format(

    'org.apache.janusgraph.some_spark_gremlin_connector',

).options(

    url='url',

    query='g.V().hasLabel("user").has("age", gt(29)).out("buy").hasLabel("item").valueMap("item_id", "name")',

).load().dropna()

df_sale = spark_session.read.format(

    'org.apache.janusgraph.some_spark_gremlin_connector',

).options(

    url='url',

    query='g.V().hasLabel("user").has("age", gt(29)).outE("buy").valueMap("timestamp")',

).load().select(

    col('item_id'),

    col('name'),

).dropna()

--------

2017年8月18日金曜日 4時08分02秒 UTC+9 HadoopMarc:

Hi Takao,

Only some directions. If you combine:

http://yaaics.blogspot.nl/              (using CassandraInputFormat in your case)
http://tinkerpop.apache.org/docs/current/reference/#interacting-with-spark

it should be possible to access the PersistedInputRDD alias graphRDD from the Spark object.
Never done this myself, I would be interested to read if this works! Probably you will need to run an OLAP query with SparkGraphComputer anyway (e.g. g.V()) to have the PersistedInputRDD realized (RDD's are not realized until a spark action is run on them.)

Cheers,     Marc


Op donderdag 17 augustus 2017 16:25:42 UTC+2 schreef Takao Magoori:

I have a JanusGraph Server (github master, gremlin 3.2.5) on top of Cassandra storage backend, to store users, items and "WHEN, WHERE, WHO bought WHAT ?" relations.

To get data from and modify data in the graph, I use Python aiogremlin driver-mode (== groovy sessionless eval mode) and it works well for now. Thanks developers !

So now, I have to compute recommendation and forecast item sales.

In order to data-cleaning, data-normalization, recommendation and forecasting, Because of a little big graph, I want to use higher-level pyspark tools (ex. DataFrame, ML) and python machine learning packages (ex, scikit-learn). But I can not find the way to load graph data into Spark. What I want is "connector" which can be used by pyspark to load data from JanusGraph, not SparkGraphComputer.

Could someone please how to do it ?

- Additional info

It seems OrientDB has some Spark connectors (though, I don't know these can be used by pyspark). But I want JanusGraph's one.

https://github.com/sbcd90/spark-orientdb

https://github.com/metreta/spark-orientdb-connector