[
https://issues.apache.org/jira/browse/HDFS-9721?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15124472#comment-15124472
]
Xiao Chen commented on HDFS-9721:
---------------------------------
Thanks very much for reviewing [~eddyxu]! I'll address the last 2 comments
shortly.
I'm not sure about the log level though. Logging it at INFO or above will flood
the console, and may confuse the user. For example, in a big cluster, currently
the tool prints the following:
{noformat}
16/01/29 11:07:01 INFO offlineImageViewer.PBImageTextWriter: Found 44801245
INodes in the INode section
16/01/29 11:22:34 INFO offlineImageViewer.PBImageTextWriter: Ignored 2235 nodes.
16/01/29 11:22:34 INFO offlineImageViewer.PBImageTextWriter: Outputted 44801245
INodes.
{noformat}
If we log every exception, there will be 2235 log entries, hence flooding out
the summary info.
I understand your concern. How about I add a sentence to the {{Ignored 2235
nodes.}} to say {{please turn on debug log for details?}}, and change this log
level to WARN?
> Allow Delimited PB OIV tool to run upon fsimage that contains INodeReference
> ----------------------------------------------------------------------------
>
> Key: HDFS-9721
> URL: https://issues.apache.org/jira/browse/HDFS-9721
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Xiao Chen
> Assignee: Xiao Chen
> Attachments: HDFS-9721.01.patch, HDFS-9721.02.patch,
> HDFS-9721.03.patch
>
>
> HDFS-6673 added the feature of Delimited format OIV tool on protocol buffer
> based fsimage.
> However, if the fsimage contains {{INodeReference}}, the tool fails because:
> {code}Preconditions.checkState(e.getRefChildrenCount() == 0);{code}
> This jira is to propose allow the tool to finish, so that user can get full
> metadata.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
