[TAJO-1872] Increase the minimum split size and add a classpath to hadoop tools

• Type: Task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.11.0
• Component/s: conf and scripts
• Assignee: Jihoon Son
• Reporter: Jihoon Son
• Created: 21/Sep/15 03:48
• Updated: 21/Sep/15 06:35
• Resolved: 21/Sep/15 06:06

Description

The current minimum split size is 1 byte. This can cause a wrong behaviour when the underlying file system gives the wrong block size.
For example, s3a file system sometimes returns 0, so tajo creates a split for each byte as follows.

default> \d customer

table name: default.customer
table uri: s3a://id:key@tajo-data-sa-east-1/tpch-1g/customer/customer.tbl
store type: TEXT
number of rows: unknown
volume: 24.3 MB

...
2015-09-21 12:42:37,806 INFO org.apache.tajo.storage.FileTablespace: Total # of splits: 24346144
...