[
https://issues.apache.org/jira/browse/HDFS-9598?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15117906#comment-15117906
]
Colin Patrick McCabe commented on HDFS-9598:
--------------------------------------------
Hmm. It looks like the code already has a fallback.
{code}
/**
* Allocate a direct buffer of the specified size, in bytes.
* If a pooled buffer is available, returns that. Otherwise
* allocates a new one.
*/
public ByteBuffer getBuffer(int size) {
Queue<WeakReference<ByteBuffer>> list = buffersBySize.get(size);
if (list == null) {
// no available buffers for this size
return ByteBuffer.allocateDirect(size);
}
WeakReference<ByteBuffer> ref;
while ((ref = list.poll()) != null) {
ByteBuffer b = ref.get();
if (b != null) {
return b;
}
}
return ByteBuffer.allocateDirect(size);
}
{code}
It sounds like what is happening here is that we failed to allocate a direct
byte buffer. I'm not sure what we should do in this case...
> Use heap buffer in PacketReceiver If hdfs client has no enough free direct
> buffer.
> ----------------------------------------------------------------------------------
>
> Key: HDFS-9598
> URL: https://issues.apache.org/jira/browse/HDFS-9598
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Yong Zhang
> Assignee: Yong Zhang
> Attachments: HDFS-9598.001.patch, HDFS-9598.002.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
