[
https://issues.apache.org/jira/browse/FLINK-7499?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16264394#comment-16264394
]
ASF GitHub Bot commented on FLINK-7499:
---------------------------------------
Github user NicoK commented on a diff in the pull request:
https://github.com/apache/flink/pull/4581#discussion_r152813605
--- Diff:
flink-runtime/src/test/java/org/apache/flink/runtime/io/network/partition/SpillableSubpartitionTest.java
---
@@ -192,21 +198,39 @@ public void testConsumeSpilledPartition() throws
Exception {
Buffer read = reader.getNextBuffer();
assertNotNull(read);
+ assertNotSame(buffer, read);
+ assertFalse(read.isRecycled());
read.recycle();
+ assertTrue(read.isRecycled());
read = reader.getNextBuffer();
assertNotNull(read);
+ assertNotSame(buffer, read);
+ assertFalse(read.isRecycled());
read.recycle();
+ assertTrue(read.isRecycled());
read = reader.getNextBuffer();
assertNotNull(read);
+ assertNotSame(buffer, read);
+ assertFalse(read.isRecycled());
read.recycle();
+ assertTrue(read.isRecycled());
// End of partition
read = reader.getNextBuffer();
assertNotNull(read);
assertEquals(EndOfPartitionEvent.class,
EventSerializer.fromBuffer(read,
ClassLoader.getSystemClassLoader()).getClass());
+ assertFalse(read.isRecycled());
read.recycle();
+ assertTrue(read.isRecycled());
+
+ // finally check that the buffer has been freed after a
successful (or failed) write
+ final long deadline = System.currentTimeMillis() + 30_000L; //
30 secs
+ while (!buffer.isRecycled() && System.currentTimeMillis() <
deadline) {
+ Thread.sleep(1);
+ }
+ assertTrue(buffer.isRecycled());
--- End diff --
No, it's not recycled in `partition.releaseMemory()` directly (or at least
should not! - which is fixed now). The buffer will be recycled by the
asynchronous writer thread once its content has been written to disk, i.e.
after the `partition.releaseMemory()` call - I included such a check right
before that call but actually, this is of limited use.
> double buffer release in SpillableSubpartitionView
> --------------------------------------------------
>
> Key: FLINK-7499
> URL: https://issues.apache.org/jira/browse/FLINK-7499
> Project: Flink
> Issue Type: Sub-task
> Components: Network
> Affects Versions: 1.2.0, 1.1.4, 1.3.0, 1.1.5, 1.2.1, 1.2.2, 1.3.1, 1.4.0,
> 1.3.2, 1.3.3
> Reporter: Nico Kruber
> Assignee: Nico Kruber
> Priority: Blocker
>
> {{SpillableSubpartitionView#releaseMemory()}} recycles its buffers twice:
> once asynchronously after the write operation and once in
> {{SpillableSubpartitionView#releaseMemory()}} after adding the write
> operation to the queue.
> 1) if {{SpillableSubpartitionView#releaseMemory()}} hits first and the buffer
> is recycled, the memory region may already be reused despite the pending write
> 2) If, for some reason (probably only in tests like
> {{SpillableSubpartitionTest#testConsumeSpillablePartitionSpilledDuringConsume()}}?),
> the buffer is retained and to be used in parallel somewhere else it may also
> not be available anymore or contain corrupt data.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
