[
https://issues.apache.org/jira/browse/SPARK-6378?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Zhaokang Wang updated SPARK-6378:
---------------------------------
Attachment: TripletsViewDonotUpdate.scala
I have met a similar problem with triplets update in GraphX.
I think I have a code demo that can reproduce the situation of this issue.
I reproduce the issue on a small toy graph with only 3 vertices. My demo code
has been attached as [^TripletsViewDonotUpdate.scala].
Let me describe the steps to reproduce the issue:
1. We have constructed a small graph ({{purGraph}} in the code) with only 3
vertices. The edges of the graph are: 2->1, 3->1, 2->3.
2. Conduct the collect neighbors operation to get the {{inNeighborGraph}} of
the {{purGraph}}.
3. Outer join the {{inNeighborGraph}} vertices on {{purGraph}} to get the
{{dataGraph}}. In {{dataGraph}}, each vertex will store an ArrayBuffer of its
in neighbors' vertex id list.
4. Now we can examine the {{inNeighbor}} attribute in {{dataGraph.vertices}}
view and {{dataGraph.triplets}} view. We can see from the output that the two
views are inconsistent on vertex 3's {{inNeighbor}} property:
{quote}
----> dataGraph.vertices
vid: 1, inNeighbor:2,3
vid: 3, inNeighbor:2
vid: 2, inNeighbor:
----> dataGraph.triplets.srcAttr
vid: 2, inNeighbor:
vid: 2, inNeighbor:
vid: 3, inNeighbor:
{quote}
5. If we comment the {{purGraph.triplets.count()}} statement in the code, the
bug will disappear:
{code}
val purGraph = Graph(dataVertex, dataEdge).persist()
// purGraph.triplets.count() // !!!comment this
val inNeighborGraph = purGraph.collectNeighbors(EdgeDirection.In)
// Now join the in neighbor vertex id list to every vertex's property
val dataGraph = purGraph.outerJoinVertices(inNeighborGraph)((vid, property,
inNeighborList) => {
val inNeighborVertexIds = inNeighborList.getOrElse(Array[(VertexId,
VertexProperty)]()).map(t => t._1)
property.inNeighbor ++= inNeighborVertexIds.toBuffer
property
})
{code}
It seems that the triplets view and the vertex view of the same graph may be
inconsistent in some situation.
> srcAttr in graph.triplets don't update when the size of graph is huge
> ---------------------------------------------------------------------
>
> Key: SPARK-6378
> URL: https://issues.apache.org/jira/browse/SPARK-6378
> Project: Spark
> Issue Type: Bug
> Components: GraphX
> Affects Versions: 1.2.1
> Reporter: zhangzhenyue
> Attachments: TripletsViewDonotUpdate.scala
>
>
> when the size of the graph is huge(0.2 billion vertex, 6 billion edges), the
> srcAttr and dstAttr in graph.triplets don't update when using the
> Graph.outerJoinVertices(when the data in vertex is changed).
> the code and the log is as follows:
> {quote}
> g = graph.outerJoinVertices()...
> g,vertices,count()
> g.edges.count()
> println("example edge " + g.triplets.filter(e => e.srcId ==
> 5000000001L).collect()
> .map(e =>(e.srcId + ":" + e.srcAttr + ", " + e.dstId + ":" +
> e.dstAttr)).mkString("\n"))
> println("example vertex " + g.vertices.filter(e => e._1 ==
> 5000000001L).collect()
> .map(e => (e._1 + "," + e._2)).mkString("\n"))
> {quote}
> the result:
> {quote}
> example edge 5000000001:0, 2467451620:61
> 5000000001:0, 1962741310:83 // attr of vertex 5000000001 is 0 in
> Graph.triplets
> example vertex 5000000001,2 // attr of vertex 5000000001 is 2 in
> Graph.vertices
> {quote}
> when the graph is smaller(10 million vertex), the code is OK, the triplets
> will update when the vertex is changed
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
