[
https://issues.apache.org/jira/browse/HADOOP-997?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tom White updated HADOOP-997:
-----------------------------
Attachment: HADOOP-997.patch
Here's a patch for review.
It actually has a general retry mechanism, a la HADOOP-601. (There is no
annotations support, but it wouldn't be too hard to extend RetryProxy to have a
method that introspects the runtime method annotations of a passed in object
and constructs a proxy based on them.) I've put it in
org.apache.hadoop.io.retry, but I'm not sure about this as it isn't IO
specific. The idea is that this mechanism can be reused for HDFS, MapRed and
IPC retries.
On the S3 front, I have increased the block size to 64 MB (since we don't have
to worry about the whole block being buffered in memory any longer), and I've
removed the notes about JetS3t buffering, since this shouldn't need to be
changed by users now.
I haven't tried it with any big files yet...
> Implement S3 retry mechanism for failed block transfers
> -------------------------------------------------------
>
> Key: HADOOP-997
> URL: https://issues.apache.org/jira/browse/HADOOP-997
> Project: Hadoop
> Issue Type: Improvement
> Components: fs
> Affects Versions: 0.11.0
> Reporter: Tom White
> Assigned To: Tom White
> Attachments: HADOOP-997.patch
>
>
> HADOOP-882 improves S3FileSystem so that when certain communications problems
> with S3 occur the operation is retried. However, the retry mechanism cannot
> handle a block transfer failure, since blocks may be very large and we don't
> want to buffer them in memory. This improvement is to write a wrapper (using
> java.lang.reflect.Proxy if possible - see discussion in HADOOP-882) that can
> retry block transfers.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
