[
https://issues.apache.org/jira/browse/HDFS-11598?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15951830#comment-15951830
]
Wei-Chiu Chuang commented on HDFS-11598:
----------------------------------------
Hello [~linyiqun] thanks for your quick patch!
This is probably my oversight initially....
If we do not add an erasure coded file into waitList, we do not need to check
if a file is erasure coded in waitForReplication(), correct?
For the error message "Do not set replication for: ", it reads more naturally
to use past tense "Did not set replication for:"
Also in the test, would you mind to add an additional assert to make sure the
replication factor of the EC file is still 1?
> Improve -setrep for Erasure Coded files
> ---------------------------------------
>
> Key: HDFS-11598
> URL: https://issues.apache.org/jira/browse/HDFS-11598
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: shell
> Affects Versions: 3.0.0-alpha1
> Reporter: Wei-Chiu Chuang
> Assignee: Yiqun Lin
> Labels: hdfs-ec-3.0-nice-to-have
> Attachments: HDFS-11598.001.patch
>
>
> Quoting my comments in HDFS-10974:
> {quote}
> Looking at code again, I think we can improve the handling of EC files better:
> * In SetReplication#processPath, can we check if the file is EC before
> invoking setReplication() on it?
> * -setrep also has a command switch -w to wait for the file to be replicated.
> Can we ignore this if the file is erasure coded?
> {quote}
> Also, in fact, -setrep only make sense for HDFS file system...
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
