[SPARK-36614][CORE][UI] Correct executor loss reason caused by decommission in UI
### What changes were proposed in this pull request? Post the correct executor loss reason to UI. ### Why are the changes needed? To show the accurate loss reason. ### Does this PR introduce _any_ user-facing change? Yes. Users can see the difference from the UI. Before: <img width="509" alt="WeChataad8d1f27d9f9aa7cf93ced4bcc820e2" src="https://user-images.githubusercontent.com/16397174/131341692-6f412607-87b8-405e-822d-0d28f07928da.png"> <img width="1138" alt="WeChat13c9f1345a096ff83d193e4e9853b165" src="https://user-images.githubusercontent.com/16397174/131341699-f2c9de09-635f-49df-8e27-2495f34276c0.png"> After: <img width="599" alt="WeChata4313fa2dbf27bf2dbfaef5c1d4a19cf" src="https://user-images.githubusercontent.com/16397174/131341754-e3c93b5d-5252-4006-a4cc-94d76f41303b.png"> <img width="1182" alt="WeChat5559d52fd3070ae6c42fe32d56f9dc94" src="https://user-images.githubusercontent.com/16397174/131341761-e1e0644f-1e76-49c0-915a-26aad77ec272.png"> ### How was this patch tested? Manully tested. Closes #33868 from Ngone51/fix-executor-remove-reason. Authored-by: yi.wu <yi.wu@databricks.com> Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
This commit is contained in:
parent
fcc91cfec4
commit
ebe7bb6217
|
@ -429,8 +429,8 @@ class CoarseGrainedSchedulerBackend(scheduler: TaskSchedulerImpl, val rpcEnv: Rp
|
|||
totalCoreCount.addAndGet(-executorInfo.totalCores)
|
||||
totalRegisteredExecutors.addAndGet(-1)
|
||||
scheduler.executorLost(executorId, lossReason)
|
||||
listenerBus.post(
|
||||
SparkListenerExecutorRemoved(System.currentTimeMillis(), executorId, reason.toString))
|
||||
listenerBus.post(SparkListenerExecutorRemoved(
|
||||
System.currentTimeMillis(), executorId, lossReason.toString))
|
||||
case None =>
|
||||
// SPARK-15262: If an executor is still alive even after the scheduler has removed
|
||||
// its metadata, we may receive a heartbeat from that executor and tell its block
|
||||
|
|
Loading…
Reference in a new issue