Github user staple commented on a diff in the pull request:
https://github.com/apache/spark/pull/1362#discussion_r15701858
--- Diff:
core/src/test/scala/org/apache/spark/scheduler/DAGSchedulerSuite.scala ---
@@ -291,6 +293,18 @@ class DAGSchedulerSuite extends
TestKit(ActorSystem("DAGSchedulerSuite")) with F
assertDataStructuresEmpty
}
+ test("avoid exponential blowup when getting preferred locs list") {
+ // Build up a complex dependency graph with repeated zip operations,
without preferred locations.
+ var rdd: RDD[_] = new MyRDD(sc, 1, Nil)
+ (1 to 30).foreach(_ => rdd = rdd.zip(rdd))
+ // getPreferredLocs runs quickly, indicating that exponential graph
traversal is avoided.
+ failAfter(5 seconds) {
--- End diff --
Sure, changed to 10 seconds. The old DAGScheduler took about 4 minutes on
my system.
---
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.
---
