[
https://issues.apache.org/jira/browse/HDFS-14546?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16872374#comment-16872374
]
Ayush Saxena commented on HDFS-14546:
-------------------------------------
Thanx [~Amithsha] for the update.
* The patch doesn't apply to trunk, Read
[https://cwiki.apache.org/confluence/display/HADOOP/How+To+Contribute] on how
to create patch. If still issues let me know, Will try to help. :)
* Guess forgot to remove the old one :
{code:java}
+https://issues.apache.org/jira/browse/HADOOP-8468
+For more information check [HADOOP-8468](
https://issues.apache.org/jira/browse/HADOOP-8468){code}
* Seems like messed up here, for the JIRA link points to the doc and the JIRA
link is as it is. Once you flip it correct. Can have a cover for DOC link also.
A Line and [Some Text to act as
link]([https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HdfsUpgradeDomain.html])
and you need to give relative path instead of full.
{code:java}
+https://issues.apache.org/jira/browse/HDFS-9006
+For more information check
[HDFS-8131](https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HdfsUpgradeDomain.html)
{code}
* For {{AvailableSpaceBlockPlacementPolicy}} can even mention that this
optimization is for second replica onwards, if data locality is set to be
preserved. If it is to be preserved, it will select the local node straight
away as it is done in case of default BPP and will perform the optimization in
choosing the replicas thereafter.(Frame it in some proper sentence.)
* Should even mention about the configurations like you did for others.
{{dfs.namenode.available-space-block-placement-policy.balanced-space-preference-fraction}}
is one of which.
* For {{BlockPlacementPolicyRackFaultTolerant}} After this line : With the
BlockPlacementPolicyRackFaultTolerant we can place replicas to more than 2
racks.
You can add some more like. The policy tries to occupy the maximum possible
number of racks.
More than two might not convey the whole meaning. if the rep is 5 and racks
available is more than or equal . It would try 5.
Anyone else too giving a look?
> Document block placement policies
> ---------------------------------
>
> Key: HDFS-14546
> URL: https://issues.apache.org/jira/browse/HDFS-14546
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Íñigo Goiri
> Assignee: Amithsha
> Priority: Major
> Labels: documentation
> Attachments: HDFS-14546-01.patch, HDFS-14546-02.patch,
> HdfsDesign.patch
>
>
> Currently, all the documentation refers to the default block placement policy.
> However, over time there have been new policies:
> * BlockPlacementPolicyRackFaultTolerant (HDFS-7891)
> * BlockPlacementPolicyWithNodeGroup (HDFS-3601)
> * BlockPlacementPolicyWithUpgradeDomain (HDFS-9006)
> We should update the documentation to refer to them explaining their
> particularities and probably how to setup each one of them.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
