[
https://issues.apache.org/jira/browse/HDFS-3702?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15191996#comment-15191996
]
stack commented on HDFS-3702:
-----------------------------
bq. I am also curious about the answer to Devaraj's question. HDFS-2576 was
added specifically for HBase. Can it address your use case? This avoids any
changes to HDFS.
[~arpiagariu]
On the [~devaraj] question on why not HDFS-2576 comment from near on three
years ago, the 'favored nodes' feature was never fully-plumbed in HBase so no
one to my knowledge ever used it. While there are rumors that our brothers and
sisters at Y! are in the process of reviving it, the original implementors of
'favored nodes', FB, now consider it a 'mistake' [1] and state they'll
"...have a party when FB no longer has this operational nightmare. " Given this
report, hbase community would be wary of going a 'favored nodes' route.
IIUC, to make use of it in this case, the 'client' would have to have a
NN-like-awareness of cluster members and pick placement as the NN would
excluding localhost? It seems like a lot to ask of the client/user of dfsclient.
1.
https://issues.apache.org/jira/browse/HBASE-6721?focusedCommentId=14720273&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14720273
> Add an option for NOT writing the blocks locally if there is a datanode on
> the same box as the client
> -----------------------------------------------------------------------------------------------------
>
> Key: HDFS-3702
> URL: https://issues.apache.org/jira/browse/HDFS-3702
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: hdfs-client
> Affects Versions: 2.5.1
> Reporter: Nicolas Liochon
> Assignee: Lei (Eddy) Xu
> Priority: Minor
> Labels: BB2015-05-TBR
> Attachments: HDFS-3702.000.patch, HDFS-3702.001.patch,
> HDFS-3702.002.patch, HDFS-3702.003.patch, HDFS-3702.004.patch,
> HDFS-3702.005.patch, HDFS-3702.006.patch, HDFS-3702.007.patch,
> HDFS-3702.008.patch, HDFS-3702_Design.pdf
>
>
> This is useful for Write-Ahead-Logs: these files are writen for recovery
> only, and are not read when there are no failures.
> Taking HBase as an example, these files will be read only if the process that
> wrote them (the 'HBase regionserver') dies. This will likely come from a
> hardware failure, hence the corresponding datanode will be dead as well. So
> we're writing 3 replicas, but in reality only 2 of them are really useful.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
