[ 
https://issues.apache.org/jira/browse/HDDS-16641?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18126686#comment-18126686
 ] 

rajesh commented on HDDS-16641:
-------------------------------

Closing this after discussion with [~Sammi] .

Findings:
* \{{OmLifecycleScanState}} already supports resuming. \{{LifecycleActionTask}} 
saves the scan position periodically (state save interval / max keys per state) 
and also when the scan is aborted (service disabled, suspended, or OM no longer 
leader). So an interrupted bucket scan already resumes from the last saved 
position instead of starting over.
* The scenario this Jira targets, doing an extra save scan state call when the 
task hits \{{ozone.lifecycle.service.timeout}}, never happens today. The 
timeout is not enforced anywhere. In \{{BackgroundService}}, 
\{{serviceTimeoutInNanos}} is only used to log a warning after the task has 
already finished; tasks are never cancelled or interrupted. This applies to all 
\{{BackgroundService}} subclasses (KeyDeletingService, 
DirectoryDeletingService, OpenKeyCleanupService, etc.), not just 
KeyLifecycleService. Those services keep a run short with per-task work limits 
rather than the timeout.

So there is nothing for this Jira to do in the current state.

Follow-ups, if pursued, will be tracked separately:
* Honouring the timeout in KeyLifecycleService (stop the scan once the timeout 
is exceeded, save scan state, resume next run): a new Jira.
* Persisting scan state for the multipart upload abort scan: a known TODO.

> Make sure same lifecycle task can be resumed next time when task is timeout
> ---------------------------------------------------------------------------
>
>                 Key: HDDS-16641
>                 URL: https://issues.apache.org/jira/browse/HDDS-16641
>             Project: Apache Ozone
>          Issue Type: Sub-task
>            Reporter: Sammi Chen
>            Assignee: rajesh
>            Priority: Major
>
> Make sure same lifecycle task can be resumed next time when task was timeout. 



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to