[
https://issues.apache.org/jira/browse/HBASE-8702?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13677363#comment-13677363
]
Jesse Yates commented on HBASE-8702:
------------------------------------
My inclination here is to not support a custom WAL codec when doing
replication, particularly for 0.94. It just adds gobs of complexity from the
'regular' case of supporting custom serialziation to the WAL. In 0.96, the HLog
writing and replication serialization are completely orthogonal, even using
different serialization mechanisms, lending credence to the idea that we should
consider them separate in 0.94.
We would just need a big, fat disclaimer that the codec is only for the HLog,
not for replication (which most people probably won't even consider as being
used, at least off the bat).
> Make WALEditCodec pluggable
> ---------------------------
>
> Key: HBASE-8702
> URL: https://issues.apache.org/jira/browse/HBASE-8702
> Project: HBase
> Issue Type: Improvement
> Components: Replication, wal
> Reporter: Jesse Yates
> Assignee: Jesse Yates
> Fix For: 0.98.0, 0.95.1, 0.94.9
>
>
> WALEditCode needs to be pluggable to support alternative serialziation
> mechanisms.
> The open question here is whether to support the alternative codec when doing
> replication - both clusters would need the codec on the classpath, which has
> additional overhead and also will be a little bit complicated when making the
> WAL serialization backwards compatible in 0.94.
> This is the follow-up to HBASE-8636.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
