[
https://issues.apache.org/jira/browse/FLINK-31610?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17709317#comment-17709317
]
Weijie Guo commented on FLINK-31610:
------------------------------------
Thanks [~fanrui] for the clarification.
{noformat}
General principle: new buffers can be requested when
"numOfRequestedMemorySegments + numberOfRequestedOverdraftMemorySegments <
poolSize + maxOverdraftBuffersPerGate".
{noformat}
Totally agreed with this principle. If the previous behavior was not what it
was designed for, I will create a ticket to do this(i.e. convert
{{numberOfRequestedMemorySegments}} to
{{numberOfRequestedOverdraftMemorySegments}} when poolSize is decreased).
{noformat}
When numberOfRequestedMemorySegments <= poolSize, all buffers are ordinary
buffer.
When numberOfRequestedMemorySegments > poolSize, the `ordinary buffer size =
poolSize`, and `the overdraft buffer size = numberOfRequestedMemorySegments -
poolSize`
{noformat}
UUIC, That's exactly what I said in my first comment: {{changes the definition
of overdraft buffer from static to dynamic(i.e. Whether it is an overdraft
buffer depends on the poolSize at the current moment.)}}., This approach avoid
a lot of complexity, I'm +1 for this.
> Refactoring of LocalBufferPool
> ------------------------------
>
> Key: FLINK-31610
> URL: https://issues.apache.org/jira/browse/FLINK-31610
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Network
> Affects Versions: 1.17.0
> Reporter: Anton Kalashnikov
> Priority: Major
>
> FLINK-31293 bug highlighted the issue with the internal mutual consistency of
> different fields in LocalBufferPool. ex.:
> - `numberOfRequestedOverdraftMemorySegments`
> - `numberOfRequestedMemorySegments`
> - `availableMemorySegment`
> - `currentPoolSize`
> Most of the problem was fixed already(I hope) but it is a good idea to
> reorganize the code in such a way that all invariants between all fields
> inside will be clearly determined and difficult to break.
> As one example I can propose getting rid of
> numberOfRequestedOverdraftMemorySegments and using existing
> numberOfRequestedMemorySegments instead. That means:
> - the pool will be available when `!availableMemorySegments.isEmpty() &&
> unavailableSubpartitionsCount == 0`
> - we don't request a new `ordinary` buffer when
> `numberOfRequestedMemorySegments >= currentPoolSize` but we request the
> overdraft buffer instead
> - `setNumBuffers` should work automatically without any changes
> I think we can come up with a couple of such improvements to simplify the
> code.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
