DSXchange

Is it repeatable, or did it happen just once? If just once, it may have been a glitch in the network.

You've answered your own question.

If you don't know what the original count is, and need to know, then the only way is to interrogate the table.

You could do this in a previous job with a query such as

There *should* be an SQL tab in the stage where you can view the SQL (I don't have DataStage access today - it's the weekend). If not, it's usually logged. With identity insert ON, all you have to do is to send rows that do not contain a value for the key column; Sybase will make use of the fact tha...

Verify the output of the stage variable. For example, did you need the third element to be ".00" rather than "00"? Did you need a space between the date portion and the time portion?

I took some time to play with it, to learn its idiosyncracies. Its very value is in how primitive an operator it is. It IS worth learning for all those little things (null handling, column name change, data type change) that you often have to do to get downstream stages to work properly.

You've got the character positions misaligned. For example the month portion should be TrimD2[6,2], not TrimD2[5,2]. You didn't allow for the delimiter character.

Of course.

Just think about the numbers. The largest address you can address with a signed 32-bit number is (2^31 - 1).

That's where the 2GB limit comes from, not from any operating system limits.

Experiment with the user name. UniVerse can be finicky; it may need the user name to be qualified by the machine name or the domain name, depending upon where it's authenticated. There is a system table called UV.LOGINS on Windows machines where you can map the three variants to a single user. And, ...

The ODBC driver for UniVerse is on your UniVerse clients CD. You need to do a small amount of work on the server side to enable the UniVerse account for ODBC access. Ideally you will have an account with clean dictionaries, rather than fifteen years worth of accreted entries that seemed like a good ...

Even that was too detailed. <soapbox> Rows/sec is meaningless, especially in small jobs. Row sizes vary, the clock keeps running even when no rows are being processed, the time is rounded to whole seconds, job designs are different, and so on. Prefer something like MB/min as a throughput metric, and...

Nothing should be in quotes. Create another output link from the Transformer stage containing the values from the stage variables and the value from the derivation. Run this into, say, a Peek stage; inspect the results and diagnose. You may discover (for example) that LastName is truly empty in some...

The Last Ran timestamp in Director comes from RT_STATUSnnn table, from the job record therein. Here, nnn is the job number mapped from DS_JOBS, but you can not do that (mapping to a table name) in SQL. This is what Director has to do, but it does so programmatically, looping through the job names in...

dsrpcd listens on port #31538 for requests to use the dscs service. If you look in dsrpcservices file, you will find that dsapi_server is the executable for the dscs service. That one is running indicates that there is a connected client. Have you tried starting the DataStage telnet service in debug...

There is a suggested learning sequence mentioned at the IBM website cited.

Count the table before and after the deletion. The difference will be the number of rows deleted.