[SQOOP-1154] Sqoop2: Text partitioner might miss or include edge values

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.99.2
• Fix Version/s: 1.99.3
• Assignee: Jarek Jarcec Cecho
• Reporter: Jarek Jarcec Cecho
• Created: 30/Jul/13 03:13
• Updated: 31/Jul/13 07:25
• Resolved: 31/Jul/13 05:51

Description

Attached is test case that will use Varchar partitioner to generate 3, 5, 10 and 13 partitions for interval "Breezy Badger" to "Warty Warthog". For all cases generated partitions looks like the following:

'Bree' <= VCCOL AND VCCOL < SOME_VALUE
SOME_VALUE <= VCCOL AND VCCOL <= 'Wart'

As 'Warty Warthog' > 'Wart', the last value will be never imported. Also similarly as 'Bree' < 'Breezy Badger', additional values might be imported as well (for example 'Breedy Budget'). I think that the varchar partitioner must have the interval boundaries there without any truncation, for example:

'Breezy Badget' <= VCCOL AND VCCOL < SOME_VALUE
SOME_VALUE <= VCCOL AND VCCOL <= 'Warty Warthog'

Issue Links

• links to: Review board