[
https://issues.apache.org/jira/browse/HIVE-8267?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Muhammad Ehsan ul Haque updated HIVE-8267:
------------------------------------------
Status: Patch Available (was: Open)
Patch available.
Unable to put a review request on review-board, as the patch failing to be
uploaded. I am new to the review board.
Feature docuentation. Perhaps I should update the page
https://cwiki.apache.org/confluence/display/Hive/HBaseIntegration#HBaseIntegration-HiveHBaseIntegration
* A cell timestamp mapping using {{:timestamp:cf:qualifier}} must be mapped to
a {{BIGINT}} column of hive.
* A column family cells timestamp mapping using {{:timestamp:cf:}} or
{{:timestamp:cf:prefix.*}} must be mapped to a {{MAP<HIVE PRIMITIVE
TYPE,BIGINT>}} of hive.
* It is not allowed to insert only timestamp without a cell value. Use
{{"hbase.put.default.cell.value" = "default value"}} in the {{SERDEPROPERTIES}}
to use a default cell value if the cell value is not mapped or may have a null
value.
* Inserting with a lower timestamp then the current latest timestamp of the
cell will be inserted as an old version.
* If cell value and timestamp are both mapped and timestamp field is {{null}},
then it is filled with {{SERDEPROPERTIES}} {{hbase.put.timestamp}} if provided
otherwise it will be filled with hbase current timestamp.
> Exposing hbase cell latest timestamp through hbase columns mappings to hive
> columns.
> ------------------------------------------------------------------------------------
>
> Key: HIVE-8267
> URL: https://issues.apache.org/jira/browse/HIVE-8267
> Project: Hive
> Issue Type: Bug
> Components: HBase Handler
> Affects Versions: 0.14.0
> Reporter: Muhammad Ehsan ul Haque
> Priority: Minor
> Fix For: 0.14.0
>
> Attachments: HIVE-8267.0.patch
>
>
> Previous attempts HIVE-2781 (not accepted), HIVE-2828 (broken and proposed
> with restricted feature).
> The feature is to have hbase cell latest timestamp accessible in hive query,
> by mapping the cell timestamp with a hive column, using mapping format like
> {code}:timestamp:cf:[optional qualifier or qualifier prefix]{code}
> The hive create table statement would be like
> h4. For mapping a cell latest timestamp.
> {code}
> CREATE TABLE hive_hbase_table (key STRING, col1 STRING, col1_ts BIGINT)
> STORED BY 'org.apache.hadoop.hive.hbase.HBaseStorageHandler'
> WITH SERDEPROPERTIES ("hbase.columns.mapping" = ":key,cf:qualifier,
> :timestamp:cf:qualifier")
> TBLPROPERTIES ("hbase.table.name" = "hbase_table");
> {code}
> h4. For mapping a column family latest timestamp.
> {code}
> CREATE TABLE hive_hbase_table (key STRING, valuemap MAP<STRING, STRING>,
> timestampmap MAP<STRING, BIGINT>)
> STORED BY 'org.apache.hadoop.hive.hbase.HBaseStorageHandler'
> WITH SERDEPROPERTIES ("hbase.columns.mapping" = ":key,cf:,:timestamp:cf:")
> TBLPROPERTIES ("hbase.table.name" = "hbase_table");
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
