Author: Michael Malak <michaelmalak@yahoo.com>
Closes#5464 from michaelmalak/master and squashes the following commits:
9d942ba [Michael Malak] SPARK-6710 GraphX Fixed Wrong initial bias in GraphX SVDPlusPlus
https://issues.apache.org/jira/browse/SPARK-6758
I am not sure if it is ok to block them in test resources too (as we shade jetty in assembly?).
Author: WangTaoTheTonic <wangtao111@huawei.com>
Closes#5406 from WangTaoTheTonic/SPARK-6758 and squashes the following commits:
e09605b [WangTaoTheTonic] block the right jetty package
So we can turn style checker on for test code.
Author: Reynold Xin <rxin@databricks.com>
Closes#5410 from rxin/test-style-graphx and squashes the following commits:
89e253a [Reynold Xin] [SPARK-6765] Fix test code style for graphx.
Example of Graph#aggregateMessages has error.
Since aggregateMessages is a method of Graph, It should be written "rawGraph.aggregateMessages"
Author: Sasaki Toru <sasakitoa@nttdata.co.jp>
Closes#5388 from sasakitoa/aggregateMessagesExample and squashes the following commits:
b1d631b [Sasaki Toru] Example of Graph#aggregateMessages has error
This builds on my earlier pull requests and turns on the explicit type checking in scalastyle.
Author: Reynold Xin <rxin@databricks.com>
Closes#5342 from rxin/SPARK-6428 and squashes the following commits:
7b531ab [Reynold Xin] import ordering
2d9a8a5 [Reynold Xin] jl
e668b1c [Reynold Xin] override
9b9e119 [Reynold Xin] Parenthesis.
82e0cf5 [Reynold Xin] [SPARK-6428] Turn on explicit type checking for public methods.
Adds a `Graph#minus` method which will return only unique `VertexId`'s from the calling `VertexRDD`.
To demonstrate a basic example with pseudocode:
```
Set((0L,0),(1L,1)).minus(Set((1L,1),(2L,2)))
> Set((0L,0))
```
Author: Brennon York <brennon.york@capitalone.com>
Closes#5175 from brennonyork/SPARK-6510 and squashes the following commits:
248d5c8 [Brennon York] added minus(VertexRDD[VD]) method to avoid createUsingIndex and updated the mask operations to simplify with andNot call
3fb7cce [Brennon York] updated graphx doc to reflect the addition of minus method
6575d92 [Brennon York] updated mima exclude
aaa030b [Brennon York] completed graph#minus functionality
7227c0f [Brennon York] beginning work on minus functionality
Correct some typos. Correct a mistake in lib/PageRank.scala. The first PageRank implementation uses standalone Graph interface, but the second uses Pregel interface. It may mislead the code viewers.
Author: Hangchen Yu <yuhc@gitcafe.com>
Closes#5128 from yuhc/master and squashes the following commits:
53e5432 [Hangchen Yu] Merge branch 'master' of https://github.com/yuhc/spark
67b77b5 [Hangchen Yu] [SPARK-6455] [docs] Correct some mistakes and typos
206f2dc [Hangchen Yu] Correct some mistakes and typos.
Use `Utils.createTempDir()` to replace other temp file mechanisms used in some tests, to further ensure they are cleaned up, and simplify
Author: Sean Owen <sowen@cloudera.com>
Closes#5029 from srowen/SPARK-6338 and squashes the following commits:
27b740a [Sean Owen] Fix hive-thriftserver tests that don't expect an existing dir
4a212fa [Sean Owen] Standardize a bit more temp dir management
9004081 [Sean Owen] Revert some added recursive-delete calls
57609e4 [Sean Owen] Use Utils.createTempDir() to replace other temp file mechanisms used in some tests, to further ensure they are cleaned up, and simplify
This extractor is mainly used for Graph#aggregateMessages*.
Author: Takeshi YAMAMURO <linguin.m.s@gmail.com>
Closes#5047 from maropu/AddUnapplyInEdgeContext and squashes the following commits:
87e04df [Takeshi YAMAMURO] Add unapply in EdgeContext
Changed method invocation of 'diff' to match that of 'innerJoin' and 'leftJoin' from VertexRDD[VD] to RDD[(VertexId, VD)]. This change maintains backwards compatibility and better unifies the VertexRDD methods to match each other.
Author: Brennon York <brennon.york@capitalone.com>
Closes#4733 from brennonyork/SPARK-5922 and squashes the following commits:
e800f08 [Brennon York] fixed merge conflicts
b9274af [Brennon York] fixed merge conflicts
f86375c [Brennon York] fixed minor include line
398ddb4 [Brennon York] fixed merge conflicts
aac1810 [Brennon York] updated to aggregateUsingIndex and added test to ensure that method works properly
2af0b88 [Brennon York] removed deprecation line
753c963 [Brennon York] fixed merge conflicts and set preference to use the diff(other: VertexRDD[VD]) method
2c678c6 [Brennon York] added mima exclude to exclude new public diff method from VertexRDD
93186f3 [Brennon York] added back the original diff method to sustain binary compatibility
f18356e [Brennon York] changed method invocation of 'diff' to match that of 'innerJoin' and 'leftJoin' from VertexRDD[VD] to RDD[(VertexId, VD)]
Added tests that maropu [created](1f64794b2c/graphx/src/test/scala/org/apache/spark/graphx/VertexRDDSuite.scala) for vertices with differing partition counts. Wanted to make sure his work got captured /merged as its not in the master branch and I don't believe there's a PR out already for it.
Author: Brennon York <brennon.york@capitalone.com>
Closes#5023 from brennonyork/SPARK-5790 and squashes the following commits:
83bbd29 [Brennon York] added maropu's tests for vertices with differing partition counts
Turns out, per the [convo on the JIRA](https://issues.apache.org/jira/browse/SPARK-4600), `diff` is acting exactly as should. It became a large misconception as I thought it meant set difference, when in fact it does not. To that extent I merely updated the `diff` documentation to, hopefully, better reflect its true intentions moving forward.
Author: Brennon York <brennon.york@capitalone.com>
Closes#5015 from brennonyork/SPARK-4600 and squashes the following commits:
1e1d1e5 [Brennon York] reverted internal diff docs
92288f7 [Brennon York] reverted both the test suite and the diff function back to its origin functionality
f428623 [Brennon York] updated diff documentation to better represent its function
cc16d65 [Brennon York] Merge remote-tracking branch 'upstream/master' into SPARK-4600
66818b9 [Brennon York] added small secondary diff test
99ad412 [Brennon York] Merge remote-tracking branch 'upstream/master' into SPARK-4600
74b8c95 [Brennon York] corrected method by leveraging bitmask operations to correctly return only the portions of that are different from the calling VertexRDD
9717120 [Brennon York] updated diff impl to cause fewer objects to be created
710a21c [Brennon York] working diff given test case
aa57f83 [Brennon York] updated to set ShortestPaths to run 'forward' rather than 'backward'
The issue is discussed in https://issues.apache.org/jira/browse/SPARK-5669. Replacing all JBLAS usage by netlib-java gives us a simpler dependency tree and less license issues to worry about. I didn't touch the test scope in this PR. The user guide is not modified to avoid merge conflicts with branch-1.3. srowen ankurdave pwendell
Author: Xiangrui Meng <meng@databricks.com>
Closes#4699 from mengxr/SPARK-5814 and squashes the following commits:
48635c6 [Xiangrui Meng] move netlib-java version to parent pom
ca21c74 [Xiangrui Meng] remove jblas from ml-guide
5f7767a [Xiangrui Meng] Merge remote-tracking branch 'apache/master' into SPARK-5814
c5c4183 [Xiangrui Meng] merge master
0f20cad [Xiangrui Meng] add mima excludes
e53e9f4 [Xiangrui Meng] remove jblas from mllib runtime
ceaa14d [Xiangrui Meng] replace jblas by netlib-java in graphx
fa7c2ca [Xiangrui Meng] move jblas to test scope
Option 1 of 2: Convert spark-parent module name to spark-parent_2.10 / spark-parent_2.11
Author: Sean Owen <sowen@cloudera.com>
Closes#4912 from srowen/SPARK-6182.1 and squashes the following commits:
eff60de [Sean Owen] Convert spark-parent module name to spark-parent_2.10 / spark-parent_2.11
Class TaskContext is unused in EdgeRDDImpl, so we need to remove it from import list.
Author: Lianhui Wang <lianhuiwang09@gmail.com>
Closes#4846 from lianhuiwang/SPARK-6103 and squashes the following commits:
31aed64 [Lianhui Wang] remove unused class to import in EdgeRDDImpl
Fixes the issue whereby when VertexRDD's are `diff`ed, `innerJoin`ed, or `leftJoin`ed and have different partition sizes they fail under the `zipPartitions` method. This fix tests whether the partitions are equal or not and, if not, will repartition the other to match the partition size of the calling VertexRDD.
Author: Brennon York <brennon.york@capitalone.com>
Closes#4705 from brennonyork/SPARK-1955 and squashes the following commits:
0882590 [Brennon York] updated to properly handle differently-partitioned vertexRDDs
Now, deprecated runSVDPlusPlus and update run, for 1.4.0 / master only
Author: Sean Owen <sowen@cloudera.com>
Closes#4625 from srowen/SPARK-5815.2 and squashes the following commits:
6fd2ca5 [Sean Owen] Now, deprecated runSVDPlusPlus and update run, for 1.4.0 / master only
Deprecate SVDPlusPlus.run and introduce SVDPlusPlus.runSVDPlusPlus with return type that doesn't include DoubleMatrix
CC mengxr
Author: Sean Owen <sowen@cloudera.com>
Closes#4614 from srowen/SPARK-5815 and squashes the following commits:
288cb05 [Sean Owen] Clarify deprecation plans in scaladoc
497458e [Sean Owen] Deprecate SVDPlusPlus.run and introduce SVDPlusPlus.runSVDPlusPlus with return type that doesn't include DoubleMatrix
This just unpersist()s each RDD in this code that was cache()ed.
Author: Sean Owen <sowen@cloudera.com>
Closes#4234 from srowen/SPARK-3290 and squashes the following commits:
66c1e11 [Sean Owen] unpersist() each RDD that was cache()ed
Corrected the logic with ShortestPaths so that the calculation will run forward rather than backwards. Output before looked like:
```scala
import org.apache.spark.graphx._
val g = Graph(sc.makeRDD(Array((1L,""), (2L,""), (3L,""))), sc.makeRDD(Array(Edge(1L,2L,""), Edge(2L,3L,""))))
lib.ShortestPaths.run(g,Array(3)).vertices.collect
// res0: Array[(org.apache.spark.graphx.VertexId, org.apache.spark.graphx.lib.ShortestPaths.SPMap)] = Array((1,Map()), (3,Map(3 -> 0)), (2,Map()))
lib.ShortestPaths.run(g,Array(1)).vertices.collect
// res1: Array[(org.apache.spark.graphx.VertexId, org.apache.spark.graphx.lib.ShortestPaths.SPMap)] = Array((1,Map(1 -> 0)), (3,Map(1 -> 2)), (2,Map(1 -> 1)))
```
And new output after the changes looks like:
```scala
import org.apache.spark.graphx._
val g = Graph(sc.makeRDD(Array((1L,""), (2L,""), (3L,""))), sc.makeRDD(Array(Edge(1L,2L,""), Edge(2L,3L,""))))
lib.ShortestPaths.run(g,Array(3)).vertices.collect
// res0: Array[(org.apache.spark.graphx.VertexId, org.apache.spark.graphx.lib.ShortestPaths.SPMap)] = Array((1,Map(3 -> 2)), (2,Map(3 -> 1)), (3,Map(3 -> 0)))
lib.ShortestPaths.run(g,Array(1)).vertices.collect
// res1: Array[(org.apache.spark.graphx.VertexId, org.apache.spark.graphx.lib.ShortestPaths.SPMap)] = Array((1,Map(1 -> 0)), (2,Map()), (3,Map()))
```
Author: Brennon York <brennon.york@capitalone.com>
Closes#4478 from brennonyork/SPARK-5343 and squashes the following commits:
aa57f83 [Brennon York] updated to set ShortestPaths to run 'forward' rather than 'backward'
When I build a graph with a file format error, there will be an ArrayIndexOutOfBoundsException
Author: Leolh <leosandylh@gmail.com>
Closes#4176 from Leolh/patch-1 and squashes the following commits:
94f6d22 [Leolh] Update GraphLoader.scala
23767f1 [Leolh] [SPARK-3650][GraphX] There will be an ArrayIndexOutOfBoundsException if the format of the source file is wrong
Try to redesign the "primitive type => Writable" implicit APIs to make them be activated automatically and without breaking binary compatibility.
However, this PR will breaking the source compatibility if people use `xxxToXxxWritable` occasionally. See the unit test in `graphx`.
Author: zsxwing <zsxwing@gmail.com>
Closes#3642 from zsxwing/SPARK-4795 and squashes the following commits:
914b2d6 [zsxwing] Add implicit back to the Writables methods
0b9017f [zsxwing] Add some docs
a0e8509 [zsxwing] Merge branch 'master' into SPARK-4795
39343de [zsxwing] Fix the unit test
64853af [zsxwing] Reorganize the rest 'implicit' methods in SparkContext
This fixes getStorageLevel for EdgeRDDImpl and VertexRDDImpl (and therefore for Graph).
See code example on JIRA which failed before but works with this patch: [https://issues.apache.org/jira/browse/SPARK-5534]
(The added unit tests also failed before but work with this fix.)
Note: I used partitionsRDD, assuming that getStorageLevel will only be called on the driver.
CC: mengxr (related to LDA PR), rxin ankurdave Thanks in advance!
Author: Joseph K. Bradley <joseph@databricks.com>
Closes#4317 from jkbradley/graphx-storagelevel and squashes the following commits:
1c21e49 [Joseph K. Bradley] made graph getStorageLevel test more robust
18d64ca [Joseph K. Bradley] Added tests for getStorageLevel in VertexRDDSuite, EdgeRDDSuite, GraphSuite
17b488b [Joseph K. Bradley] overrode getStorageLevel in Vertex/EdgeRDDImpl to use partitionsRDD
Added the 2 methods to Graph and GraphImpl. Both make calls to the underlying vertex and edge RDDs.
This is needed for another PR (for LDA): [https://github.com/apache/spark/pull/4047]
Notes:
* getCheckpointedFiles is plural and returns a Seq[String] instead of an Option[String].
* I attempted to test to make sure the methods returned the correct values after checkpointing. It did not work; I guess that checkpointing does not occur quickly enough? I noticed that there are not checkpointing tests for RDDs; is it just hard to test well?
CC: rxin
CC: mengxr (since related to LDA)
Author: Joseph K. Bradley <joseph@databricks.com>
Closes#4253 from jkbradley/graphx-checkpoint and squashes the following commits:
b680148 [Joseph K. Bradley] added class tag to firstParent call in VertexRDDImpl.isCheckpointed, though not needed to compile
250810e [Joseph K. Bradley] In EdgeRDDImple, VertexRDDImpl, added transient back to partitionsRDD, and made isCheckpointed check firstParent instead of partitionsRDD
695b7a3 [Joseph K. Bradley] changed partitionsRDD in EdgeRDDImpl, VertexRDDImpl to be non-transient
cc00767 [Joseph K. Bradley] added overrides for isCheckpointed, getCheckpointFile in EdgeRDDImpl, VertexRDDImpl. The corresponding Graph methods now work.
188665f [Joseph K. Bradley] improved documentation
235738c [Joseph K. Bradley] Added isCheckpointed and getCheckpointFiles to Graph, GraphImpl
These are more `javadoc` 8-related changes I spotted while investigating. These should be helpful in any event, but this does not nearly resolve SPARK-3359, which may never be feasible while using `unidoc` and `javadoc` 8.
Author: Sean Owen <sowen@cloudera.com>
Closes#4193 from srowen/SPARK-3359 and squashes the following commits:
5b33f66 [Sean Owen] Additional scaladoc fixes for javadoc 8; still not going to be javadoc 8 compatible
One side-effect of shading guava is that it disappears as a transitive
dependency. For Hadoop 2.x, this was masked by the fact that Hadoop
itself depends on guava. But certain versions of Hadoop 1.x also
shade guava, leaving either no guava or some random version pulled
by another dependency on the classpath.
So be explicit about the dependency in modules that use guava directly,
which is the right thing to do anyway.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#4272 from vanzin/SPARK-5466 and squashes the following commits:
e3f30e5 [Marcelo Vanzin] Dependency for catalyst is not needed.
d3b2c84 [Marcelo Vanzin] [SPARK-5466] Add explicit guava dependencies where needed.
If the value of 'spark.default.parallelism' does not match the number of partitoins in EdgePartition(EdgeRDDImpl),
the following error occurs in ReplicatedVertexView.scala:72;
object GraphTest extends Logging {
def run[VD: ClassTag, ED: ClassTag](graph: Graph[VD, ED]): VertexRDD[Int] = {
graph.aggregateMessages(
ctx => {
ctx.sendToSrc(1)
ctx.sendToDst(2)
},
_ + _)
}
}
val g = GraphLoader.edgeListFile(sc, "graph.txt")
val rdd = GraphTest.run(g)
java.lang.IllegalArgumentException: Can't zip RDDs with unequal numbers of partitions
at org.apache.spark.rdd.ZippedPartitionsBaseRDD.getPartitions(ZippedPartitionsRDD.scala:57)
at org.apache.spark.rdd.RDD$$anonfun$partitions$2.apply(RDD.scala:206)
at org.apache.spark.rdd.RDD$$anonfun$partitions$2.apply(RDD.scala:204)
at scala.Option.getOrElse(Option.scala:120)
at org.apache.spark.rdd.RDD.partitions(RDD.scala:204)
at org.apache.spark.rdd.MapPartitionsRDD.getPartitions(MapPartitionsRDD.scala:32)
at org.apache.spark.rdd.RDD$$anonfun$partitions$2.apply(RDD.scala:206)
at org.apache.spark.rdd.RDD$$anonfun$partitions$2.apply(RDD.scala:204)
at scala.Option.getOrElse(Option.scala:120)
at org.apache.spark.rdd.RDD.partitions(RDD.scala:204)
at org.apache.spark.ShuffleDependency.<init>(Dependency.scala:82)
at org.apache.spark.rdd.ShuffledRDD.getDependencies(ShuffledRDD.scala:80)
at org.apache.spark.rdd.RDD$$anonfun$dependencies$2.apply(RDD.scala:193)
at org.apache.spark.rdd.RDD$$anonfun$dependencies$2.apply(RDD.scala:191)
...
Author: Takeshi Yamamuro <linguin.m.s@gmail.com>
Closes#4136 from maropu/EdgePartitionBugFix and squashes the following commits:
0cd8942 [Ankur Dave] Use more concise getOrElse
aad4a2c [Ankur Dave] Add unit test for non-default number of edge partitions
0a2f32b [Takeshi Yamamuro] Do not use Partitioner.defaultPartitioner as a partitioner of EdgeRDDImpl
I looked into GraphGenerators#chooseCell, and found that chooseCell can't generate more edges than pow(2, (2 * (log2(numVertices)-1))) to make a Power-law graph. (Ex. numVertices:4 upperbound:4, numVertices:8 upperbound:16, numVertices:16 upperbound:64)
If we request more edges over the upperbound, rmatGraph fall into infinite loop. So, how about adding an argument validation?
Author: Kenji Kikushima <kikushima.kenji@lab.ntt.co.jp>
Closes#3950 from kj-ki/SPARK-5064 and squashes the following commits:
4ee18c7 [Ankur Dave] Reword error message and add unit test
d760bc7 [Kenji Kikushima] Add numEdges upperbound validation for R-MAT graph generator to prevent infinite loop.
This change does a few things to make the hadoop-provided profile more useful:
- Create new profiles for other libraries / services that might be provided by the infrastructure
- Simplify and fix the poms so that the profiles are only activated while building assemblies.
- Fix tests so that they're able to run when the profiles are activated
- Add a new env variable to be used by distributions that use these profiles to provide the runtime
classpath for Spark jobs and daemons.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#2982 from vanzin/SPARK-4048 and squashes the following commits:
82eb688 [Marcelo Vanzin] Add a comment.
eb228c0 [Marcelo Vanzin] Fix borked merge.
4e38f4e [Marcelo Vanzin] Merge branch 'master' into SPARK-4048
9ef79a3 [Marcelo Vanzin] Alternative way to propagate test classpath to child processes.
371ebee [Marcelo Vanzin] Review feedback.
52f366d [Marcelo Vanzin] Merge branch 'master' into SPARK-4048
83099fc [Marcelo Vanzin] Merge branch 'master' into SPARK-4048
7377e7b [Marcelo Vanzin] Merge branch 'master' into SPARK-4048
322f882 [Marcelo Vanzin] Fix merge fail.
f24e9e7 [Marcelo Vanzin] Merge branch 'master' into SPARK-4048
8b00b6a [Marcelo Vanzin] Merge branch 'master' into SPARK-4048
9640503 [Marcelo Vanzin] Cleanup child process log message.
115fde5 [Marcelo Vanzin] Simplify a comment (and make it consistent with another pom).
e3ab2da [Marcelo Vanzin] Fix hive-thriftserver profile.
7820d58 [Marcelo Vanzin] Fix CliSuite with provided profiles.
1be73d4 [Marcelo Vanzin] Restore flume-provided profile.
d1399ed [Marcelo Vanzin] Restore jetty dependency.
82a54b9 [Marcelo Vanzin] Remove unused profile.
5c54a25 [Marcelo Vanzin] Fix HiveThriftServer2Suite with *-provided profiles.
1fc4d0b [Marcelo Vanzin] Update dependencies for hive-thriftserver.
f7b3bbe [Marcelo Vanzin] Add snappy to hadoop-provided list.
9e4e001 [Marcelo Vanzin] Remove duplicate hive profile.
d928d62 [Marcelo Vanzin] Redirect child stderr to parent's log.
4d67469 [Marcelo Vanzin] Propagate SPARK_DIST_CLASSPATH on Yarn.
417d90e [Marcelo Vanzin] Introduce "SPARK_DIST_CLASSPATH".
2f95f0d [Marcelo Vanzin] Propagate classpath to child processes during testing.
1adf91c [Marcelo Vanzin] Re-enable maven-install-plugin for a few projects.
284dda6 [Marcelo Vanzin] Rework the "hadoop-provided" profile, add new ones.
Convert bi-directional edges into uni-directional ones instead of 'canonicalOrientation' in GraphLoader.edgeListFile.
This function is useful when a graph is loaded as it is and then is transformed into one with canonical edges.
It rewrites the vertex ids of edges so that srcIds are bigger than dstIds, and merges the duplicated edges.
Author: Takeshi Yamamuro <linguin.m.s@gmail.com>
Closes#3760 from maropu/ConvertToCanonicalEdgesSpike and squashes the following commits:
7f8b580 [Takeshi Yamamuro] Add a function to convert into a graph with canonical edges in GraphOps
This PR:
- Reenables `surefire`, and copies config from `scalatest` (which is itself an old fork of `surefire`, so similar)
- Tells `surefire` to test only Java tests
- Enables `surefire` and `scalatest` for all children, and in turn eliminates some duplication.
For me this causes the Scala and Java tests to be run once each, it seems, as desired. It doesn't affect the SBT build but works for Maven. I still need to verify that all of the Scala tests and Java tests are being run.
Author: Sean Owen <sowen@cloudera.com>
Closes#3651 from srowen/SPARK-4159 and squashes the following commits:
2e8a0af [Sean Owen] Remove specialized SPARK_HOME setting for REPL, YARN tests as it appears to be obsolete
12e4558 [Sean Owen] Append to unit-test.log instead of overwriting, so that both surefire and scalatest output is preserved. Also standardize/correct comments a bit.
e6f8601 [Sean Owen] Reenable Java tests by reenabling surefire with config cloned from scalatest; centralize test config in the parent
The sum of vertices on matrix (v0 to v11) is 12. And, I think one same block overlaps in this strategy.
This is minor PR, so I didn't file in JIRA.
Author: kj-ki <kikushima.kenji@lab.ntt.co.jp>
Closes#3904 from kj-ki/fix-partitionstrategy-comments and squashes the following commits:
79829d9 [kj-ki] Fix comments for 2D partitioning.
As we learned in #3580, not explicitly typing implicit functions can lead to compiler bugs and potentially unexpected runtime behavior.
This is a follow up PR for rest of Spark (outside Spark SQL). The original PR for Spark SQL can be found at https://github.com/apache/spark/pull/3859
Author: Reynold Xin <rxin@databricks.com>
Closes#3860 from rxin/implicit and squashes the following commits:
73702f9 [Reynold Xin] [SPARK-5038] Add explicit return type for implicit functions.
Add an IF to uncache both vertices and edges of Graph/GraphImpl.
This IF is useful when iterative graph operations build a new graph in each iteration, and the vertices and edges of previous iterations are no longer needed for following iterations.
Author: Takeshi Yamamuro <linguin.m.s@gmail.com>
This patch had conflicts when merged, resolved by
Committer: Ankur Dave <ankurdave@gmail.com>
Closes#3476 from maropu/UnpersistInGraphSpike and squashes the following commits:
77a006a [Takeshi Yamamuro] Add unpersist in Graph and GraphImpl
This patch just replaces a native quick sorter with Sorter(TimSort) in Spark.
It could get performance gains by ~8% in my quick experiments.
Author: Takeshi Yamamuro <linguin.m.s@gmail.com>
Closes#3507 from maropu/TimSortInEdgePartitionBuilderSpike and squashes the following commits:
8d4e5d2 [Takeshi Yamamuro] Remove a wildcard import
3527e00 [Takeshi Yamamuro] Replace Scala.util.Sorting.quickSort with Sorter(TimSort) in Spark
Author: GuoQiang Li <witgo@qq.com>
Closes#2631 from witgo/SPARK-3623 and squashes the following commits:
a70c500 [GuoQiang Li] Remove java related
4d1e249 [GuoQiang Li] Add comments
e682724 [GuoQiang Li] Graph should support the checkpoint operation
The related JIRA is https://issues.apache.org/jira/browse/SPARK-4672
In a nutshell, if `val partitionsRDD` in EdgeRDDImpl and VertexRDDImpl are non-transient, the serialization chain can become very long in iterative algorithms and finally lead to the StackOverflow error. More details and explanation can be found in the JIRA.
Author: JerryLead <JerryLead@163.com>
Author: Lijie Xu <csxulijie@gmail.com>
Closes#3544 from JerryLead/my_graphX and squashes the following commits:
628f33c [JerryLead] set PartitionsRDD to be transient in EdgeRDDImpl and VertexRDDImpl
c0169da [JerryLead] Merge branch 'master' of https://github.com/apache/spark
52799e3 [Lijie Xu] Merge pull request #1 from apache/master
The related JIRA is https://issues.apache.org/jira/browse/SPARK-4672
Iterative GraphX applications always have long lineage, while checkpoint() on EdgeRDD and VertexRDD themselves cannot shorten the lineage. In contrast, if we perform checkpoint() on their ParitionsRDD, the long lineage can be cut off. Moreover, the existing operations such as cache() in this code is performed on the PartitionsRDD, so checkpoint() should do the same way. More details and explanation can be found in the JIRA.
Author: JerryLead <JerryLead@163.com>
Author: Lijie Xu <csxulijie@gmail.com>
Closes#3549 from JerryLead/my_graphX_checkpoint and squashes the following commits:
d1aa8d8 [JerryLead] Perform checkpoint() on PartitionsRDD not VertexRDD and EdgeRDD themselves
ff08ed4 [JerryLead] Merge branch 'master' of https://github.com/apache/spark
c0169da [JerryLead] Merge branch 'master' of https://github.com/apache/spark
52799e3 [Lijie Xu] Merge pull request #1 from apache/master
After additional discussion with rxin, I think having all the possible `TripletField` options is confusing. This pull request reduces the triplet fields to:
```java
/**
* None of the triplet fields are exposed.
*/
public static final TripletFields None = new TripletFields(false, false, false);
/**
* Expose only the edge field and not the source or destination field.
*/
public static final TripletFields EdgeOnly = new TripletFields(false, false, true);
/**
* Expose the source and edge fields but not the destination field. (Same as Src)
*/
public static final TripletFields Src = new TripletFields(true, false, true);
/**
* Expose the destination and edge fields but not the source field. (Same as Dst)
*/
public static final TripletFields Dst = new TripletFields(false, true, true);
/**
* Expose all the fields (source, edge, and destination).
*/
public static final TripletFields All = new TripletFields(true, true, true);
```
Author: Joseph E. Gonzalez <joseph.e.gonzalez@gmail.com>
Closes#3472 from jegonzal/SimplifyTripletFields and squashes the following commits:
91796b5 [Joseph E. Gonzalez] removing confusing triplet fields
This pull request revises the programming guide to reflect changes in the GraphX API as well as the deprecated mapReduceTriplets operator.
Author: Joseph E. Gonzalez <joseph.e.gonzalez@gmail.com>
Closes#3359 from jegonzal/GraphXProgrammingGuide and squashes the following commits:
4421964 [Joseph E. Gonzalez] updating documentation for graphx
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#3277 from vanzin/version-1.3 and squashes the following commits:
7c3c396 [Marcelo Vanzin] Added temp repo to sbt build.
5f404ff [Marcelo Vanzin] Add another exclusion.
19457e7 [Marcelo Vanzin] Update old version to 1.2, add temporary 1.2 repo.
3c8d705 [Marcelo Vanzin] Workaround for MIMA checks.
e940810 [Marcelo Vanzin] Bumping version to 1.3.0-SNAPSHOT.
Due to vertex attribute caching, EdgeRDD previously took two type parameters: ED and VD. However, this is an implementation detail that should not be exposed in the interface, so this PR drops the VD type parameter.
This requires removing the `filter` method from the EdgeRDD interface, because it depends on vertex attribute caching.
Author: Ankur Dave <ankurdave@gmail.com>
Closes#3303 from ankurdave/edgerdd-drop-tparam and squashes the following commits:
38dca9b [Ankur Dave] Leave EdgeRDD.fromEdges public
fafeb51 [Ankur Dave] Drop VD type parameter from EdgeRDD
This discourages users from calling the VertexRDD and EdgeRDD constructor and makes it easier for future changes to ensure backward compatibility.
Author: Ankur Dave <ankurdave@gmail.com>
Closes#2530 from ankurdave/SPARK-3666 and squashes the following commits:
d681f45 [Ankur Dave] Define getPartitions and compute in abstract class for MIMA
1472390 [Ankur Dave] Merge remote-tracking branch 'apache-spark/master' into SPARK-3666
24201d4 [Ankur Dave] Merge remote-tracking branch 'apache-spark/master' into SPARK-3666
cbe15f2 [Ankur Dave] Remove specialized annotation from VertexRDD and EdgeRDD
931b587 [Ankur Dave] Use abstract class instead of trait for binary compatibility
9ba4ec4 [Ankur Dave] Mark (Vertex|Edge)RDDImpl constructors package-private
620e603 [Ankur Dave] Extract VertexRDD interface and move implementation to VertexRDDImpl
55b6398 [Ankur Dave] Extract EdgeRDD interface and move implementation to EdgeRDDImpl
1. Add EdgeActiveness enum to represent activeness criteria more cleanly than using booleans.
2. Comments and whitespace.
Author: Ankur Dave <ankurdave@gmail.com>
Closes#3231 from ankurdave/aggregateMessages-followup and squashes the following commits:
3d485c3 [Ankur Dave] Internal cleanup for aggregateMessages
aggregateMessages enables neighborhood computation similarly to mapReduceTriplets, but it introduces two API improvements:
1. Messages are sent using an imperative interface based on EdgeContext rather than by returning an iterator of messages.
2. Rather than attempting bytecode inspection, the required triplet fields must be explicitly specified by the user by passing a TripletFields object. This fixes SPARK-3936.
Additionally, this PR includes the following optimizations for aggregateMessages and EdgePartition:
1. EdgePartition now stores local vertex ids instead of global ids. This avoids hash lookups when looking up vertex attributes and aggregating messages.
2. Internal iterators in aggregateMessages are inlined into a while loop.
In total, these optimizations were tested to provide a 37% speedup on PageRank (uk-2007-05 graph, 10 iterations, 16 r3.2xlarge machines, sped up from 513 s to 322 s).
Subsumes apache/spark#2815. Also fixes SPARK-4173.
Author: Ankur Dave <ankurdave@gmail.com>
Closes#3100 from ankurdave/aggregateMessages and squashes the following commits:
f5b65d0 [Ankur Dave] Address @rxin comments on apache/spark#3054 and apache/spark#3100
1e80aca [Ankur Dave] Add aggregateMessages, which supersedes mapReduceTriplets
194a2df [Ankur Dave] Test triplet iterator in EdgePartition serialization test
e0f8ecc [Ankur Dave] Take activeSet in ExistingEdgePartitionBuilder
c85076d [Ankur Dave] Readability improvements
b567be2 [Ankur Dave] iter.foreach -> while loop
4a566dc [Ankur Dave] Optimizations for mapReduceTriplets and EdgePartition
As [reported][1] on the mailing list, GraphX throws
```
java.lang.ClassCastException: java.lang.Long cannot be cast to scala.Tuple2
at org.apache.spark.graphx.impl.RoutingTableMessageSerializer$$anon$1$$anon$2.writeObject(Serializers.scala:39)
at org.apache.spark.storage.DiskBlockObjectWriter.write(BlockObjectWriter.scala:195)
at org.apache.spark.util.collection.ExternalSorter.spillToMergeableFile(ExternalSorter.scala:329)
```
when sort-based shuffle attempts to spill to disk. This is because GraphX defines custom serializers for shuffling pair RDDs that assume Spark will always serialize the entire pair object rather than breaking it up into its components. However, the spill code path in sort-based shuffle [violates this assumption][2].
GraphX uses the custom serializers to compress vertex ID keys using variable-length integer encoding. However, since the serializer can no longer rely on the key and value being serialized and deserialized together, performing such encoding would either require writing a tag byte (costly) or maintaining state in the serializer and assuming that serialization calls will alternate between key and value (fragile).
Instead, this PR simply removes the custom serializers. This causes a **10% slowdown** (494 s to 543 s) and **16% increase in per-iteration communication** (2176 MB to 2518 MB) for PageRank (averages across 3 trials, 10 iterations per trial, uk-2007-05 graph, 16 r3.2xlarge nodes).
[1]: http://apache-spark-user-list.1001560.n3.nabble.com/java-lang-ClassCastException-java-lang-Long-cannot-be-cast-to-scala-Tuple2-td13926.html#a14501
[2]: f9d6220c79/core/src/main/scala/org/apache/spark/util/collection/ExternalSorter.scala (L329)
Author: Ankur Dave <ankurdave@gmail.com>
Closes#2503 from ankurdave/SPARK-3649 and squashes the following commits:
a49c2ad [Ankur Dave] [SPARK-3649] Remove GraphX custom serializers
at first srcIds is not initialized and are all 0. so we use edgeArray(0).srcId to currSrcId
Author: lianhuiwang <lianhuiwang09@gmail.com>
Closes#3138 from lianhuiwang/SPARK-4249 and squashes the following commits:
3f4e503 [lianhuiwang] fix a problem of EdgePartitionBuilder in Graphx
