[
https://issues.apache.org/jira/browse/HADOOP-6949?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12986137#action_12986137
]
Doug Cutting commented on HADOOP-6949:
--------------------------------------
We don't currently support cross-version RPC, so changing the RPC wire format
is certainly acceptable. My concern and the reason I suggested adding a
subclass of ObjectWritable used by RPC that writes the new format is that folks
might be using ObjectWritable for file-based data. In this case files written
by the new version would not be readable by the old version. Some sites run
different versions of Hadoop on different clusters and use distcp/HFTP to sync
sets of files between the clusters. If these files contained ObjectWritable
then this could make files sync'd from a cluster running a newer version
unreadable on a cluster running an older version. So the safest change would
be to only change RPC and not ObjectWritable.
> Reduces RPC packet size for primitive arrays, especially long[], which is
> used at block reporting
> -------------------------------------------------------------------------------------------------
>
> Key: HADOOP-6949
> URL: https://issues.apache.org/jira/browse/HADOOP-6949
> Project: Hadoop Common
> Issue Type: Improvement
> Components: io
> Reporter: Navis
> Assignee: Matt Foley
> Fix For: 0.23.0
>
> Attachments: arrayprim.patch, ObjectWritable.diff
>
> Original Estimate: 0.17h
> Remaining Estimate: 0.17h
>
> Current implementation of oah.io.ObjectWritable marshals primitive array
> types as general object array ; array type string + array length + (element
> type string + value)*n
> It would not be needed to specify each element types for primitive arrays.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
