DSXchange

Hi All,
trying to run column analysis on a file +-700MB
OS: RHL 6.2
IA: 11.5
Virtual Server with 8 CPUs
RAM: 32GB

getting the error next in line to the additional info in the topic:
pxbridge(2),0: com.ascential.e2.common.CC_Exception: java.lang.IllegalArgumentException: The value of -472844 for field Scale is either too small or too large to fit into the SmallInt type.
at com.ibm.is.cc.javastage.connector.CC_JavaRecordImpl.checkValue(CC_JavaRecordImpl.java:637)
at com.ibm.is.cc.javastage.connector.CC_JavaRecordImpl.setValue(CC_JavaRecordImpl.java:162)
at com.ibm.infosphere.ia.columnanalysis.operators.ColumnAnalysisResultsAggregationOperator.process(ColumnAnalysisResultsAggregationOperator.java:299)
at com.ibm.is.cc.javastage.connector.CC_JavaAdapter.run(CC_JavaAdapter.java:443)
at com.ibm.is.cc.javastage.connector.CC_JavaAdapter.run (CC_JavaAdapter.java: 458)
this is on the IA automatic OSH jobs for column analysis
Any idea?
(while everyone is on Xmas holiday lol )

So far I started analysis of each aborted column individualy to find the column that aborts; its only 1 column.

Also foound in the first warning:
"pxbridge(2): Unable to determine association between statement parameters and table columns. The connector will not be able to obtain external schema and only limited schema reconciliation will be performed"
but all is automatic, examining the data shows up to 20 characters and the default 255 characters were used as default metadata for profiling
next significant notice:
"pxbridge(4): [Input link 0] Unable to determine association between statement parameters and table columns. The connector will not be able to obtain external schema and only limited schema reconciliation will be performed"
This post informs that the warnings are not related

Thanks in advance...

What is the source of the data? -472844 is certainly not a smallint, so it looks like the declared datatype in the source metadata may be wrong (or you may have data corruption..)

I suspect either a bug or misuse of the product.
No definition of any kind was set except a string of 255 characters to all columns; the idea is that IA will suggest the type.

Specific analysis for that column only with 50K sample on random with 100% was successful

Hi Roy,

Indeed - a PMR may be necessary.

yes Robert,
this is caused when trying to analyse a string like "123456E3456789"
having this format causes the infered data type to be decimal
but the number of the characters after the 'E' is to long for a smallint
E is being translated as an exponent
if we analyse "1234A234323452E453635747" there is no problem
since the "A" negates the posibility of a decimal number

Regards from Alex Lavrov
A PMR has been opened...

Did you get this resolved? I have just opened up a PMR for myself. We are having the identical issue.

Domenic

For the record, it looks like this was fixed in the Governance Rollup 1 that was released 01 March.