DSXchange

How many nodes do you have in your configuration file and do you repartition?

C'mon Saik - if one of the values has implied decimal places then you need to adjust for that, either by multiplying or dividing and truncating on of the values so that they have the same scale. You can compare integers with decimal/float for '>', '<' easily when they are the same scale. Comparing f...

You can force a comparison of types by ensuring that both are converted to a common format. If you don't have leading zeroes, then a INT(In.DecimalColumn)=In.IntegerColumn should be enough.

A dataset consists of a descriptor (the file you actually refer to in the job) which contains such things as the schema layout and links to the physical data files. You can move the descriptor file around at will, it is very small. You cannot move the actual data files around without modifying your ...

I think that this link makes a good start; it is to IBM/Ascential's site and touches upon several of the questions you asked.

The problem has now been solved. It turns out that the NFS mounts on the DB/2 machines for PX had different access rights, so PX and server could connect to the main coordinator node and issue "normal" work since the main DB/2 coordinator node then went to all of the actually processing (partition) ...

Thanks, we've gone through all that already and know about that part. It seems that we are now getting connectivity to the main DB/2 coordinator node (since we can connect & read data using the user SQL) but once we parallel and use the TABLE method the connections don't go through and return -1...

We have an EE configuration that goes to 2 remote DB/2 servers. We have just completed the installation and setup of a new series of machines and have hit an odd problem. Using the DB/2 Enterprise stage and "user SQL" or "Auto generated SQL" we can view data and run a job. But when the "Table" metho...

I am a bit confused - your error message talks about a column called "MessageId" which you haven't listed in your sample data. Also, what have you defined your output as (fixed length? Quoted columns?)?

It means character 245 from the beginning of the line (or file in this case). If you have multibyte NLS then it means character 245. If you look at your column definitions and add the lengths, this is probably the position 1 of the column starting after your error column.

Yes. Use an editor and verify that line 1 is longer in this column (if delimited) or longer in total (if not delimited). This would be offset 245 from the beginning of the line.

A hashed file key is unique. If you declared 2 key columns then, in order to do a lookup, both of those columns must contain values. So if you have 2 key columns you need 2 values in order to read a record.

the RESIZE * * * will show any data errors in the file, as you surmised. It also has the benefit in this case of making sure that he is the only person accessing the file (i.e. required to rebuild indices)

A) What is the index status reported by "LIST.INDEX DS_JOBS ALL"
B) What happens when you try "RESIZE DS_JOBS * * *" or "FIX.FILE DS_JOBS"?

That might work in your case because it is the install; when DS bombs out the core will always point to uvsh, though.