[
https://issues.apache.org/jira/browse/PHOENIX-2030?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Rajeshbabu Chintaguntla updated PHOENIX-2030:
---------------------------------------------
Description:
Currently the CsvBulkLoad tool loads the data parrallely to data and index
tables. If a table has multiple local indexes then the output directory for the
indexes data is will be same and then one thread might delete when it's job
done and then other index data loading fail with FileNotFoundException.
To avoid this we should use logical name as suffix output directory which is
unique for data table and index tables.
was:
Currently the CsvBulkLoad tool loads the data parrallely to data and index
tables. If a table has multiple local indexes then the output directory for the
indexes data is will be same and then one thread might delete when it's job
done and then other index data loading fail with FileNotFoundException.
To avoid this we should use logical name as suffix output directory.
> CsvBulkLoadTool should use logical name of the table for output directory
> suffix
> --------------------------------------------------------------------------------
>
> Key: PHOENIX-2030
> URL: https://issues.apache.org/jira/browse/PHOENIX-2030
> Project: Phoenix
> Issue Type: Bug
> Reporter: Rajeshbabu Chintaguntla
> Assignee: Rajeshbabu Chintaguntla
>
> Currently the CsvBulkLoad tool loads the data parrallely to data and index
> tables. If a table has multiple local indexes then the output directory for
> the indexes data is will be same and then one thread might delete when it's
> job done and then other index data loading fail with FileNotFoundException.
> To avoid this we should use logical name as suffix output directory which is
> unique for data table and index tables.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
