[
https://issues.apache.org/jira/browse/HADOOP-6356?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14006535#comment-14006535
]
Colin Patrick McCabe commented on HADOOP-6356:
----------------------------------------------
I think the performance situation may not be as bad as you think. HDFS caches
things like sockets and short-circuit file descriptors in a separate, global
cache that is not per-{{FileContext}}. They are cached in
{{org.apache.hadoop.hdfs.ClientContext}} and will be shared between multiple
different {{FileContext}} objects.
The big problems with caching {{FileContext}} objects are:
* they're mutable, so you don't know when another thread will call
{{FileContext#setWorkingDirectory}} and screw up how your thread is resolving
relative paths
* UGI has to be a part of the "lookup key" for any cache, and we have refused
to compare UGIs except by object equality in the past. This can lead to (for
some) counter-intuitive cache behavior.
Given these limitations, I think it's better to have the applications do their
own caching. Maybe we could provide a utility class that would help with this,
but we should stay away from adding more global variables.
> Add a Cache for AbstractFileSystem in the new FileContext/AbstractFileSystem
> framework.
> ---------------------------------------------------------------------------------------
>
> Key: HADOOP-6356
> URL: https://issues.apache.org/jira/browse/HADOOP-6356
> Project: Hadoop Common
> Issue Type: Improvement
> Components: fs
> Affects Versions: 0.22.0
> Reporter: Sanjay Radia
> Assignee: Sanjay Radia
>
> The new filesystem framework, FileContext and AbstractFileSystem does not
> implement a cache for AbstractFileSystem.
> This Jira proposes to add a cache to the new framework just like with the old
> FileSystem.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
