[
https://issues.apache.org/jira/browse/HDFS-4849?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13689443#comment-13689443
]
Sanjay Radia commented on HDFS-4849:
------------------------------------
bq. Suresh, I still don't see what is wrong with the patch I provided
Konstantine this is deeper than a patch issue. Both Todd and Suresh have
brought up the multi-threaded issue. The counter argument is that a
multi-threaded client should not use the same FileSystem object; I don't buy
this argument - a multi-threaded client should be able to use the same FS
object - the object is a reference to the FileSystem and not to a particular
open file. Besides there are multi-threaded applications sharing the same FS
object and hence it would break them. Steve has also argued that there are
fundamental issues to be addressed and that is is not a mere patch issue.
Further, given that we will have to build a retry-cache anyway for other
operations like rename, why don't we do that and make optimizations for
create-with-overwrite.
We are all in agreement on the goal. Your optimization for create, in not using
the retry-cache, is changing semantics - it is not a mere implementation change.
> Idempotent create and append operations.
> ----------------------------------------
>
> Key: HDFS-4849
> URL: https://issues.apache.org/jira/browse/HDFS-4849
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: namenode
> Affects Versions: 2.0.4-alpha
> Reporter: Konstantin Shvachko
> Assignee: Konstantin Shvachko
> Priority: Blocker
> Attachments: idempotentCreate.patch, idempotentCreate.patch,
> idempotentCreate.patch
>
>
> create, append and delete operations can be made idempotent. This will reduce
> chances for a job or other app failures when NN fails over.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
