[
https://issues.apache.org/jira/browse/HDFS-7853?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14347370#comment-14347370
]
Zhe Zhang commented on HDFS-7853:
---------------------------------
Thanks Jing for the patch and sorry for the delayed review.
The overall structure looks good to me. The following implementation details
are worth more discussions:
# You might have already started optimizing {{BlockInfoStriped#indices}} based
on TODO msg. Just a reminder that
{{BlockInfoStripedUnderConstruction#blockIndices}} is only applicable in
over-replication as well. But if we get rid of indices for non-over-replicated
{{replicas}}, we need to update {{addReplicaIfNotPresent}} as well, to always
insert in the right position
# How about using a Map for those over-replicated replica (in addition to
triplets and the {{replicas}} array)? It's less efficient but should simplify
the code. It's a rare condition anyway.
# If the non-over-replicated locations in {{LocatedBlockStriped}} are sorted,
we can remove the required {{indices}} filed in PB and add an optional field
for the excess replicas and their indices.
> Erasure coding: extend LocatedBlocks to support reading from striped files
> --------------------------------------------------------------------------
>
> Key: HDFS-7853
> URL: https://issues.apache.org/jira/browse/HDFS-7853
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Zhe Zhang
> Assignee: Jing Zhao
> Attachments: HDFS-7853.000.patch
>
>
> We should extend {{LocatedBlocks}} class so {{getBlockLocations}} can work
> with striping layout (possibly an extra list specifying the index of each
> location in the group)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
