[ https://issues.apache.org/jira/browse/HDFS-4849?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13705281#comment-13705281 ]
Konstantin Shvachko commented on HDFS-4849: ------------------------------------------- > In Hadoop summit we agreed that this change will not make create idempotent. > You are only enabling retry I am not arguing, as we agreed. We also agreed that create retry is worth using in order to minimize the retry cache size. As an optimization. > please consider updating the jira description accordingly. Let me know how you want it to be. > How did you solve the issue of two threads in a single client calling create? As I argued before this is a common problem with retries or without. And the solution is to guarantee unique client names, which should be done in a different jira. If I understood your question correctly. > Please see my analysis of namenode operations in HDFS-4974. Very good analysis. Thanks. > I feel we should use retry cache instead of doing a separate retry support > for create alone. This is not what agreed upon at the Summit. Let me know if you changed your mind. > 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-branch2.patch, idempotentCreate.patch, > 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