[
https://issues.apache.org/jira/browse/HDFS-3513?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13423129#comment-13423129
]
Alejandro Abdelnur commented on HDFS-3513:
------------------------------------------
I've done some testing, accesing HttpFS via 'hadoop fs' and the difference with
caching is significant when doing multiple short operations. Below an example
doing a recursive LS (I've run the command multiple times in each
configuration before taking a sample).
For distcp usecases I don't think will make much difference once data is being
streamed (which should account for most of the distcp time). But for
integration from other systems that may do regular lookups I think there is
merit in having FS caching in HttpFS.
Also, the numbers below are without security enabled. I would presume that with
security ON the lag will be greater.
Without caching disabled:
{code}
$ time bin/hadoop fs -fs webhdfs://localhost:14000 -lsr /
lsr: DEPRECATED: Please use 'ls -R' instead.
drwxr-x--- - tucu supergroup 0 2012-07-26 07:47 /tmp
drwxr-x--- - tucu supergroup 0 2012-07-26 07:47 /tmp/hadoop-yarn
drwxr-x--- - tucu supergroup 0 2012-07-26 07:47
/tmp/hadoop-yarn/staging
drwxr-x--- - tucu supergroup 0 2012-07-26 07:47
/tmp/hadoop-yarn/staging/history
drwxrwx--- - tucu supergroup 0 2012-07-26 07:47
/tmp/hadoop-yarn/staging/history/done
drwxrwxrwt - tucu supergroup 0 2012-07-26 07:47
/tmp/hadoop-yarn/staging/history/done_intermediate
real 0m1.717s
user 0m1.599s
sys 0m0.105s
{code}
With caching enabled:
{code}
$ time bin/hadoop fs -fs webhdfs://localhost:14000 -lsr /
lsr: DEPRECATED: Please use 'ls -R' instead.
drwxr-x--- - tucu supergroup 0 2012-07-26 07:47 /tmp
drwxr-x--- - tucu supergroup 0 2012-07-26 07:47 /tmp/hadoop-yarn
drwxr-x--- - tucu supergroup 0 2012-07-26 07:47
/tmp/hadoop-yarn/staging
drwxr-x--- - tucu supergroup 0 2012-07-26 07:47
/tmp/hadoop-yarn/staging/history
drwxrwx--- - tucu supergroup 0 2012-07-26 07:47
/tmp/hadoop-yarn/staging/history/done
drwxrwxrwt - tucu supergroup 0 2012-07-26 07:47
/tmp/hadoop-yarn/staging/history/done_intermediate
real 0m0.879s
user 0m1.471s
sys 0m0.101s
{code}
> HttpFS should cache filesystems
> -------------------------------
>
> Key: HDFS-3513
> URL: https://issues.apache.org/jira/browse/HDFS-3513
> Project: Hadoop HDFS
> Issue Type: Improvement
> Affects Versions: 2.0.0-alpha
> Reporter: Alejandro Abdelnur
> Assignee: Alejandro Abdelnur
> Attachments: HDFS-3513.patch, HDFS-3513.patch, HDFS-3513.patch
>
>
> HttpFS opens and closes a FileSystem instance against the backend filesystem
> (typically HDFS) on every request. The FileSystem caching is not used as it
> does not have expiration/timeout and filesystem instances in there live
> forever, for long running services like HttpFS this is not a good thing as it
> would keep connections open to the NN.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
