You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "ASF GitHub Bot (Jira)" <ji...@apache.org> on 2022/06/27 07:57:00 UTC
[jira] [Updated] (FLINK-28258) Introduce ExecutionHistory to host historical executions for each execution vertex
[ https://issues.apache.org/jira/browse/FLINK-28258?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
ASF GitHub Bot updated FLINK-28258:
-----------------------------------
Labels: pull-request-available (was: )
> Introduce ExecutionHistory to host historical executions for each execution vertex
> ----------------------------------------------------------------------------------
>
> Key: FLINK-28258
> URL: https://issues.apache.org/jira/browse/FLINK-28258
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Coordination
> Reporter: Zhu Zhu
> Assignee: Zhu Zhu
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.16.0
>
>
> With speculative execution, tracking prior executions in an {{EvictingBoundedList}} does not work. This is because when using {{EvictingBoundedList}} relies on the assumption that the historical executions are added in ascending order of attempt number successively. This is no longer true if speculative execution is enabled. e.g. 3 speculative execution attempts #1, #2, #3 are running concurrently, later #3 failed, and then #1 failed, and execution attempt #2 keeps running.
> The broken assumption may result in exceptions in REST, job archiving and so on.
> We propose to introduce an {{ExecutionHistory}} to replace {{EvictingBoundedList}}. It hosts the historical executions in a {{LinkedHashMap}} with a size bound. When the map grows beyond the size bound, elements are dropped from the head of the map (FIFO order).
--
This message was sent by Atlassian Jira
(v8.20.7#820007)