[
https://issues.apache.org/jira/browse/ARROW-2696?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ASF GitHub Bot updated ARROW-2696:
----------------------------------
Labels: pull-request-available (was: )
> [JAVA] enhance AllocationListener with an onFailedAllocation() call
> -------------------------------------------------------------------
>
> Key: ARROW-2696
> URL: https://issues.apache.org/jira/browse/ARROW-2696
> Project: Apache Arrow
> Issue Type: Improvement
> Components: Java - Memory
> Reporter: Vanco Buca
> Priority: Major
> Labels: pull-request-available
>
> If an allocation fails, there might be an action taken by user code that will
> free up some memory before the allocation can be retried again. This action
> could include flushing some temporary buffers, getting rid of some
> temporarily cached data, or even spilling state to disk.
> Today's code is forced to catch out-of-memory exceptions, and then retry the
> operation at a high level. A more elegant way to do this is to have a
> callback whenever allocation failed. The callback would inform the allocator
> whether or not to retry the allocation.
> Since the allocator already has an AllocationListener, it makes sense to
> enhance the existing listener with a new method.
> {{The proposed signature is:}}
> {{ boolean onAllocationFailed(long allocationSize);}}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
