[
https://issues.apache.org/jira/browse/HDDS-2523?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16976761#comment-16976761
]
Attila Doroszlai commented on HDDS-2523:
----------------------------------------
Thanks [~szetszwo], that would be great. I also debugged it and found so far
that this happens with standalone replication type (which TestContainerMapper
is using). I have a simple fix idea, but need to test it with other cases, too.
> BufferPool.releaseBuffer may release a buffer different than the head of the
> list
> ---------------------------------------------------------------------------------
>
> Key: HDDS-2523
> URL: https://issues.apache.org/jira/browse/HDDS-2523
> Project: Hadoop Distributed Data Store
> Issue Type: Bug
> Components: Ozone Client
> Reporter: Tsz-wo Sze
> Assignee: Attila Doroszlai
> Priority: Major
>
> {code}
> //BufferPool
> public void releaseBuffer(ByteBuffer byteBuffer) {
> // always remove from head of the list and append at last
> ByteBuffer buffer = bufferList.remove(0);
> // Ensure the buffer to be removed is always at the head of the list.
> Preconditions.checkArgument(buffer.equals(byteBuffer));
> buffer.clear();
> bufferList.add(buffer);
> Preconditions.checkArgument(currentBufferIndex >= 0);
> currentBufferIndex--;
> }
> {code}
> In the code above, it expects buffer and byteBuffer are the same object, i.e.
> buffer == byteBuffer. However the precondition is checking
> buffer.equals(byteBuffer). Unfortunately, both buffer and byteBuffer have
> remaining() == 0 so that equals(..) returns true and the precondition does
> not catch the bug.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
