[
https://issues.apache.org/jira/browse/TAJO-149?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hyunsik Choi updated TAJO-149:
------------------------------
Attachment: TAJO-149.patch
I've uploaded the patch. This patch primarily eliminates QueryConf and its
write on HDFS. Instead, Tajo uses a global system configuration file stored on
system directory. I've described the system directory in more detail at
https://wiki.apache.org/tajo/TajoInternal#DirectoryHierarchy
In addition, it cleans up the below things:
* Clean up the system directory hierarchy as you can see the above link.
* Remove all code parts using TableMeta files (.meta) stored on each table
directory.
* Clean up the TajoMaster init() and start() methods.
* Clean up the succeeded event part of Query.
> Eliminate QueryConf and its file write
> --------------------------------------
>
> Key: TAJO-149
> URL: https://issues.apache.org/jira/browse/TAJO-149
> Project: Tajo
> Issue Type: Improvement
> Components: master, worker
> Reporter: Hyunsik Choi
> Assignee: Hyunsik Choi
> Fix For: 0.2-incubating
>
> Attachments: TAJO-149.patch
>
>
> QueryConf is a subclass of org.apache.hadoop.conf.Configuration. This has a
> similar mechanism of JobConf of MapReduce. It has overheads and too big to be
> distributed across a number of nodes. I've added QueryMeta to the patch of
> TAJO-144. QueryMeta contains a set of key-value pairs and is based on
> protobuf. Since QueryMeta is a wrapper class of a protobuf message and will
> contain only specific information for each query, it will be very lightweight
> and is easy to be disseminated across a number of nodes.
> In this issue, we will replace QueryConf by QueryMeta, and we will eliminate
> the part where QueryConf is written as a xml file on HDFS.
--
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
