[
https://issues.apache.org/jira/browse/HADOOP-14473?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16034683#comment-16034683
]
Rajesh Balamohan commented on HADOOP-14473:
-------------------------------------------
Since it was easier to combine this patch with HADOOP-14478, I have merged it
and posted the revised patch there.
In the revised patch, I have fixed an issue in seek() and shared the test
results as well there. Tests were run against "japan west region" end point.
{{BlobInputStream::skip()}} is more of a no-op call. Issue was related to
closing the stream and opening it again via {{store.retrieve()}} as it would
end up creating new {{BlobInputStream}}. And that would internally need
additional http call as it needs to download blob attributes internally in
{{BlobInputStream}}. This has been avoided in the patch.
I completely agree that it would be good to get the instrumentation similar to
s3a, and it was very useful. Please let me know if this could be done in
incremental tickets.
> Optimize NativeAzureFileSystem::seek for forward seeks
> ------------------------------------------------------
>
> Key: HADOOP-14473
> URL: https://issues.apache.org/jira/browse/HADOOP-14473
> Project: Hadoop Common
> Issue Type: Bug
> Components: fs/azure
> Reporter: Rajesh Balamohan
> Assignee: Rajesh Balamohan
> Attachments: HADOOP-14473-001.patch
>
>
> {{NativeAzureFileSystem::seek()}} closes and re-opens the inputstream
> irrespective of forward/backward seek. It would be beneficial to re-open the
> stream on backward seek.
> https://github.com/apache/hadoop/blob/trunk/hadoop-tools/hadoop-azure/src/main/java/org/apache/hadoop/fs/azure/NativeAzureFileSystem.java#L889
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
