Github user squito commented on a diff in the pull request:
https://github.com/apache/spark/pull/11105#discussion_r56414687
--- Diff:
core/src/test/scala/org/apache/spark/ConsistentAccumulatorsSuite.scala ---
@@ -0,0 +1,284 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark
+
+import scala.ref.WeakReference
+
+import org.scalatest.Matchers
+
+import org.apache.spark.scheduler._
+
+
+class ConsistentAccumulatorSuite extends SparkFunSuite with Matchers with
LocalSparkContext {
+ test("single partition") {
+ sc = new SparkContext("local[2]", "test")
+ val acc : Accumulator[Int] = sc.accumulator(0, consistent = true)
+
+ val a = sc.parallelize(1 to 20, 1)
+ val b = a.map{x => acc += x; x}
+ b.cache()
+ b.count()
+ acc.value should be (210)
+ }
+
+ test("adding only the first element per partition should work even if
partition is empty") {
+ sc = new SparkContext("local[2]", "test")
+ val acc: Accumulator[Int] = sc.accumulator(0, consistent = true)
+ val a = sc.parallelize(1 to 20, 30)
+ val b = a.mapPartitions{itr =>
+ acc += 1
+ itr
+ }
+ b.count()
+ acc.value should be (30)
+ }
+
+ test("shuffled (combineByKey)") {
+ sc = new SparkContext("local[2]", "test")
+ val a = sc.parallelize(1 to 40, 5)
+ val buckets = 4
+ val b = a.map{x => ((x % buckets), x)}
+ val inputs = List(b, b.repartition(10), b.partitionBy(new
HashPartitioner(5))).map(_.cache())
+ val mapSideCombines = List(true, false)
+ inputs.foreach{input =>
+ mapSideCombines.foreach{mapSideCombine =>
+ val accs = 1.to(4).map(x => sc.accumulator(0, consistent =
true)).toList
+ val raccs = 1.to(4).map(x => sc.accumulator(0, consistent =
false)).toList
+ val List(acc, acc1, acc2, acc3) = accs
+ val List(racc, racc1, racc2, racc3) = raccs
+ val c = input.combineByKey(
+ (x: Int) => {acc1 += 1; acc += 1; racc1 += 1; racc += 1; x},
+ {(a: Int, b: Int) => acc2 += 1; acc += 1; racc2 += 1; racc += 1;
(a + b)},
+ {(a: Int, b: Int) => acc3 += 1; acc += 1; racc3 += 1; racc += 1;
(a + b)},
+ new HashPartitioner(10),
+ mapSideCombine)
+ val d = input.combineByKey(
+ (x: Int) => {acc1 += 1; acc += 1; x},
+ {(a: Int, b: Int) => acc2 += 1; acc += 1; (a + b)},
+ {(a: Int, b: Int) => acc3 += 1; acc += 1; (a + b)},
+ new HashPartitioner(2),
+ mapSideCombine)
+ val e = d.map{x => acc += 1; x}
+ c.count()
+ // If our partitioner is known then we should only create
+ // one combiner for each key value. Otherwise we should
+ // create at least that many combiners.
+ if (input.partitioner.isDefined) {
+ acc1.value should be (buckets)
+ } else {
+ acc1.value should be >= (buckets)
+ }
+ if (input.partitioner.isDefined) {
+ acc2.value should be > (0)
+ } else if (mapSideCombine) {
+ acc3.value should be > (0)
+ } else {
+ acc2.value should be > (0)
+ acc3.value should be (0)
+ }
+ acc.value should be (acc1.value + acc2.value + acc3.value)
+ val oldValues = accs.map(_.value)
+ // For one action the consistent accumulators and regular should
have the same value.
+ accs.map(_.value) should be (raccs.map(_.value))
+ c.count()
+ accs.map(_.value) should be (oldValues)
+ // Executing a second _different_ aggregation should count
everything 2x
+ d.count()
+ accs.map(_.value) should be (oldValues.map(_ * 2))
+ // Computing the mapped value on top and verify new changes are
processed and old changes
+ // are note double counted.
+ val count = e.count()
+ accs.tail.map(_.value) should be (oldValues.tail.map(_ * 2))
+ acc.value should be (oldValues.head * 2 + count)
+ }
+ }
+ }
+
+ test("map + cache + first + count") {
+ sc = new SparkContext("local[2]", "test")
+ val acc : Accumulator[Int] = sc.accumulator(0, consistent = true)
+
+ val a = sc.parallelize(1 to 20, 10)
+ val b = a.map{x => acc += x; x}
+ b.cache()
+ b.first()
+ acc.value should be > (0)
+ b.collect()
+ acc.value should be (210)
+ }
+
+ test("coalesce acc on either side") {
+ sc = new SparkContext("local[2]", "test")
+ val List(acc1, acc2, acc3) = 1.to(3).map(x => sc.accumulator(0,
consistent = true)).toList
+ val a = sc.parallelize(1 to 20, 10)
+ val b = a.map{x => acc1 += x; acc2 += x; 2 * x}
+ val c = b.coalesce(2).map{x => acc1 += x; acc3 += x; x}
+ c.count()
+ acc1.value should be (630)
+ acc2.value should be (210)
+ acc3.value should be (420)
+ }
--- End diff --
I think the coalesce test needs to do partial partition reading, of both
the original partitions and the coalesced partitions, like so (this passes):
```scala
test("coalesce with partial partition reading") {
// when coalescing, one task can completely read partitions of the
input RDDs while not reading complete
// complete partitions of the post-coalesce RDD. We make sure that the
accumulator has consistent semantics
// in these cases.
sc = new SparkContext("local[2]", "test")
val List(acc1, acc2, acc3) = 1.to(3).map(x => sc.accumulator(0,
consistent = true)).toList
val a = sc.parallelize(1 to 20, 10)
val b = a.map{x => acc1 += x; acc2 += x; 2 * x}
val c = b.coalesce(2).map{x => acc1 += x; acc3 += x; x}
// we read all of partition 1 from RDD's a & b, and part of partition 2
// however, for RDD c, we don't read any of the partitions fully
// so we should get updates for 1 partition from b, and nothing from c
c.take(3) should be ((1 to 3).map(_*2).toArray)
acc1.value should be (3) // only elements 1 & 2 came from fully read
partitions
acc2.value should be (3)
acc3.value should be (0)
// now we read a few more parts:
c.take(9) should be ((1 to 9).map(_*2).toArray)
acc1.value should be (36)
acc2.value should be (36)
acc3.value should be (0)
// a couple more, this time we read 1 of c's partitions fully
c.take(15) should be ((1 to 15).map(_*2).toArray)
acc1.value should be (215)
acc2.value should be (105)
acc3.value should be (110)
// and if we read the entire data, we get the entire value
c.count()
acc1.value should be (630)
acc2.value should be (210)
acc3.value should be (420)
}
```
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
