Github user attilapiros commented on a diff in the pull request:
https://github.com/apache/spark/pull/21068#discussion_r185447393
--- Diff:
core/src/main/scala/org/apache/spark/scheduler/cluster/CoarseGrainedSchedulerBackend.scala
---
@@ -170,8 +170,7 @@ class CoarseGrainedSchedulerBackend(scheduler:
TaskSchedulerImpl, val rpcEnv: Rp
if (executorDataMap.contains(executorId)) {
executorRef.send(RegisterExecutorFailed("Duplicate executor ID:
" + executorId))
context.reply(true)
- } else if (scheduler.nodeBlacklist != null &&
- scheduler.nodeBlacklist.contains(hostname)) {
+ } else if
(scheduler.nodeBlacklistWithExpiryTimes.contains(hostname)) {
--- End diff --
I like the current solution. As on production nodeBlacklist really cannot
be null and that null came from the tests. So we adopted the tests to be more
close to the production case (and not the other way around: adapting the
production code to be prepared for the side effect of the testing).
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
