Mailing-List: contact hdfs-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: hdfs-issues@hadoop.apache.org
Date: Tue, 25 Nov 2014 15:21:12 +0000 (UTC)
From: "Daryn Sharp (JIRA)" <jira@apache.org>
To: hdfs-issues@hadoop.apache.org
Message-ID: <JIRA.12757433.1416846566000.16283.1416928872640@Atlassian.JIRA>
In-Reply-To: <JIRA.12757433.1416846566000@Atlassian.JIRA>
References: <JIRA.12757433.1416846566000@Atlassian.JIRA>
 <JIRA.12757433.1416846566642@arcas>
Subject: [jira] [Commented] (HDFS-7433) DatanodeMap lookups & DatanodeID
 hashCodes are inefficient
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HDFS-7433?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14224667#comment-14224667 ] 

Daryn Sharp commented on HDFS-7433:
-----------------------------------

My bad mentioning {{datanodeMap}} - juggling too many changes.  {{DatanodeIDs}} are added to collections in many other places, and equality checks occur often.  My more general point is mutable hashCodes are a hidden landmine which is why I filed another jira.  Dynamic computation of the xfer addr (and by extension the hash) is inefficient and generates a lot of garbage.

I'm checking out the odd test failures.  They don't appear related, at least  the xml parsing and class def not founds.

> DatanodeMap lookups & DatanodeID hashCodes are inefficient
> ----------------------------------------------------------
>
>                 Key: HDFS-7433
>                 URL: https://issues.apache.org/jira/browse/HDFS-7433
>             Project: Hadoop HDFS
>          Issue Type: Improvement
>          Components: namenode
>    Affects Versions: 2.0.0-alpha, 3.0.0
>            Reporter: Daryn Sharp
>            Assignee: Daryn Sharp
>            Priority: Critical
>         Attachments: HDFS-7433.patch
>
>
> The datanode map is currently a {{TreeMap}}.  For many thousands of datanodes, tree lookups are ~10X more expensive than a {{HashMap}}.  Insertions and removals are up to 100X more expensive.


--
This message was sent by Atlassian JIRA
(v6.3.4#6332)