hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Daryn Sharp (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HDFS-5875) Add iterator support to INodesInPath
Date Tue, 04 Feb 2014 00:25:07 GMT

     [ https://issues.apache.org/jira/browse/HDFS-5875?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Daryn Sharp updated HDFS-5875:
------------------------------

    Attachment: HDFS-5875.patch

Adds standard iterator support.  The iterator is a nested class so it can fill-in the existing
data members in IIP with minimal change (somewhat gross).  Given how subtle bugs in this code
region cause nearly un-debuggable test failures, I wanted to disturb as little code as possible.
 Overall, it's not the prettiest change, but I intend to iteratively - no pun intended - improve
it.

I removed the (what appears to be) pre-mature optimization for capturing 1 inode.  This is
used by extraneous calls that unnecessarily re-resolve the whole path just to get the last
inode.  These will soon go away when paths are resolved once and only once.

> Add iterator support to INodesInPath
> ------------------------------------
>
>                 Key: HDFS-5875
>                 URL: https://issues.apache.org/jira/browse/HDFS-5875
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: namenode
>    Affects Versions: 2.0.0-alpha, 3.0.0
>            Reporter: Daryn Sharp
>            Assignee: Daryn Sharp
>         Attachments: HDFS-5875.patch
>
>
> "Resolve as you go" inode iteration will help with the implementation of alternative
locking schemes.  It will also be the pre-cursor for resolving paths once and only once per
operation, as opposed to ~3 times/call.
> This is an incremental and compatible change for IIP that does not break existing callers.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

Mime
View raw message