[
https://issues.apache.org/jira/browse/PHOENIX-2154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14700718#comment-14700718
]
Samarth Jain commented on PHOENIX-2154:
---------------------------------------
Until and unless we have a pre-built histogram/distribution of the index row
keys, it will always be difficult to come up with the right split points/region
boundaries for the index table. A somewhat decent alternative is to have the
initial number of splits/regions/reducers equal to a small multiple of region
servers in the cluster. Of course, there is no guarantee that the initial
region splits would result in even distribution, but hopefully over time the
regions will eventually balance out.
> Failure of one mapper should not affect other mappers in MR index build
> -----------------------------------------------------------------------
>
> Key: PHOENIX-2154
> URL: https://issues.apache.org/jira/browse/PHOENIX-2154
> Project: Phoenix
> Issue Type: Bug
> Reporter: James Taylor
> Attachments: IndexTool.java
>
>
> Once a mapper in the MR index job succeeds, it should not need to be re-done
> in the event of the failure of one of the other mappers. The initial
> population of an index is based on a snapshot in time, so new rows getting
> *after* the index build has started and/or failed do not impact it.
> Also, there's a 1:1 correspondence between index rows and table rows, so
> there's really no need to dedup. However, the index rows will have a
> different row key than the data table, so I'm not sure how the HFiles are
> split. Will they potentially overlap and is this an issue?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
