[
https://issues.apache.org/jira/browse/AURORA-291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13951395#comment-13951395
]
Maxim Khutornenko commented on AURORA-291:
------------------------------------------
We might want to forgo this change in favor of a possible accuracy loss.
Keeping it on the back burner for now.
> Preserve ancestor details to improve accuracy of SLA calculations
> -----------------------------------------------------------------
>
> Key: AURORA-291
> URL: https://issues.apache.org/jira/browse/AURORA-291
> Project: Aurora
> Issue Type: Task
> Components: Scheduler
> Reporter: Maxim Khutornenko
> Assignee: Maxim Khutornenko
>
> In order to calculate ARD without job create/update impact we would need to
> track the ancestor details (i.e. in case a LOST task gets rescheduled,
> becomes PENDING and stays PENDING longer than HISTORY_MIN_RETENTION_THRESHOLD
> while the ancestor LOST task gets pruned).
> One possible solution could be converting the ancestorID in ScheduledTask to
> something like:
> {noformat}
> struct Ancestor {
> 1: string Id
> 2: ScheduleStatus status
> 3: i64: timestampMs
> }
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.2#6252)
