[MINOR] scheduler micro opts

### What changes were proposed in this pull request? Scheduler micro optimizations to speed up the scheduling loop. ### Why are the changes needed? The scheduler is single threaded and the faster we can schedule the faster a query executes. Changes to the scheduler usually adversely affect scheduling throughput which makes it hard to prototype novel changes. Getting some headroom makes experimentation faster while not hurting production. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? Existing tests. Closes apache#40464 from alkis/scheduler-micro-opts. Authored-by: Alkis Evlogimenos <[email protected]> Signed-off-by: Hyukjin Kwon <[email protected]>
nkarpov · Mar 20, 2023 · 372b1cb · 372b1cb
1 parent 708bda3
commit 372b1cb
Showing 1 changed file with 2 additions and 5 deletions.
diff --git a/core/src/main/scala/org/apache/spark/scheduler/TaskInfo.scala b/core/src/main/scala/org/apache/spark/scheduler/TaskInfo.scala
@@ -103,11 +103,8 @@ class TaskInfo(
     // finishTime should be set larger than 0, otherwise "finished" below will return false.
     assert(time > 0)
     finishTime = time
-    if (state == TaskState.FAILED) {
-      failed = true
-    } else if (state == TaskState.KILLED) {
-      killed = true
-    }
+    failed = state == TaskState.FAILED
+    killed = state == TaskState.KILLED
   }
 
   private[spark] def launchSucceeded(): Unit = {