[
https://issues.apache.org/jira/browse/HDFS-4974?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13708957#comment-13708957
]
Suresh Srinivas commented on HDFS-4974:
---------------------------------------
bq. Holding the write lock keeps the operation thread-safe (multiple concurrent
calls won't corrupt state), but not idempotent, because repeated calls yield
side effects (one copy of the data in the file vs. multiple).
What I mean by idempotent is - currently metasave appends some information to a
log file. If the lock was not held, it is possible to have
interleaved/incorrect information in metasave. That is not the case here. All
you end up having is an additional copy of metasave in the file. That should
not result in issues is what I am thinking.
bq. I suggest that we switch to opening the output file with overwrite instead
of append to make it truly idempotent. If that change is seen as too risky,
then we could apply the retry cache to this call.
I think this seems reasonable. But it is a changed behavior we need to document
as well. Would you like to take up that up?
> Analyze and add annotations to Namenode and Datanode protocol methods to
> enable retry
> -------------------------------------------------------------------------------------
>
> Key: HDFS-4974
> URL: https://issues.apache.org/jira/browse/HDFS-4974
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: ha, namenode
> Reporter: Suresh Srinivas
>
> This jira is intended for:
> # Discussing current @Idempotent annotations in HDFS protocols and adding
> that annotation where it is missing.
> # Discuss how retry should be enabled for non-idempotent requests.
> I will post the analysis of current methods in subsequent comment.
--
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
