You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Alexandre CLEMENT (Jira)" <ji...@apache.org> on 2021/07/08 07:46:00 UTC

[jira] [Updated] (SPARK-36042) [Dynamic allocation] Executor grace period (ExecutorIdleTimeout) ignored due to nulll startTime for pods in pending state

     [ https://issues.apache.org/jira/browse/SPARK-36042?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Alexandre CLEMENT updated SPARK-36042:
--------------------------------------
    Description: 
Pending executor are always timeouted due to null startTime and funtion returning true in case of exception in parsing startTime.

 

{{In class }}ExecutorPodsAllocator:

{{private def isExecutorIdleTimedOut(state: ExecutorPodState, currentTime: Long): Boolean = {}}
{{try {}}
{{ val startTime = Instant.parse(state.pod.getStatus.getStartTime).toEpochMilli()}}
{{ currentTime - startTime > executorIdleTimeout}}
{{catch {}}
{{  case _: Exception =>}}
{{  logDebug(s"Cannot get startTime of pod ${state.pod}")}}}}
{{  true}}
{{{{}}}}}

  was:
Pending executor are always timeouted due to null startTime and funtion returning true in case of exception in parsing startTime.

private def isExecutorIdleTimedOut(state: ExecutorPodState, currentTime: Long): Boolean = {
{{ try {}}
{{ val startTime = Instant.parse(state.pod.getStatus.getStartTime).toEpochMilli()}}
{{ currentTime - startTime > executorIdleTimeout}}
{{ } catch {}}
{{ case _: Exception =>}}
{{ logDebug(s"Cannot get startTime of pod ${state.pod}")}}
{{ true}}
{{ }}}
{{}}}


> [Dynamic allocation] Executor grace period (ExecutorIdleTimeout) ignored due to nulll startTime for pods in pending state
> -------------------------------------------------------------------------------------------------------------------------
>
>                 Key: SPARK-36042
>                 URL: https://issues.apache.org/jira/browse/SPARK-36042
>             Project: Spark
>          Issue Type: Bug
>          Components: Kubernetes
>    Affects Versions: 3.1.1
>         Environment: AWS EKS with dynamic allocation 
>            Reporter: Alexandre CLEMENT
>            Priority: Major
>
> Pending executor are always timeouted due to null startTime and funtion returning true in case of exception in parsing startTime.
>  
> {{In class }}ExecutorPodsAllocator:
> {{private def isExecutorIdleTimedOut(state: ExecutorPodState, currentTime: Long): Boolean = {}}
> {{try {}}
> {{ val startTime = Instant.parse(state.pod.getStatus.getStartTime).toEpochMilli()}}
> {{ currentTime - startTime > executorIdleTimeout}}
> {{catch {}}
> {{  case _: Exception =>}}
> {{  logDebug(s"Cannot get startTime of pod ${state.pod}")}}}}
> {{  true}}
> {{{{}}}}}



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org