DSXchange

Yes, if you just login and check "ulimit -a" then you get that process' information and that might not be the same as that used in DataStage. You can compare the "dsenv" file contents to see if the ulimit values are explicitly expanded in one and not the other, but best would be ...

You can't do it simply in a job sequence, since call to jobs in sequences are effectively hard-coded at compile-time. There is the command-line "dsjob -run" invocation method which you could call from within a loop in a job sequence, or you could use the DataStage API BASIC calls directly ...

Please delete your other duplicate thread before someone replies to it.

Ulimit values for your datastage process are the relevant ones. You could write a datastage sequence that calls the UNIX "ulimit -a" command and check the runtime values.

In addition to what I posted, you could use the UNIX "crypt" and "decrypt" commands to perform your obfuscation.

It is possible that the "ulimit" settings are indeed different for the DataStage user on both your machines, as the error message suggests. Have you looked at the values on the two machines?

The goods news is that the problem can be solved, but you will need to supply more information. Is the file that you are reading declared with fixed width or variable width? Are there column separator characters? If there are column separator characters, then declare your field as VarChar(64) (just ...

The CRC32 algorithm converts any sized string into a 32-bit number; thus by definition one will have different strings that map to the same CRC32. The algorithm excels in making similar strings into very different numbers. I'm surprised that you have detected collisions in your data using CRC32 and ...

So your incoming data to the transform stage after the join is col1 col2 col3 b_col2 b_col3 105 500 507 500 507 105 500 506 500 507 105 101 201 500 507 Your logic in the constraint would seem to be correct. Try to remove the right side of the "AND" and see if you get the expected 1 output ...

did you add $APT_DUMP_SCORE to see if there is an inserted sort, despite not being able to read the complete message?

Yes, that now makes sense, the use of "deploy" confused me, since that usually means to install software or the like. Are you certain that your sequence log contains no more messages regarding executing this batch? You could also put an "ECHO" command as the first line in the bat...

What do you mean in steps 3 and 4?

is "execute_command" your routine? Since your original post stated a job sequence, wouldn't "ExecSH" be sufficient?

Unfortunately, I can't think of a way to do that with different job names. If it were one job that is multi-instance then a sequece could loop through the instances, but that is not the case here. If it is any consolation, I have a sequence right now that has over 60 different jobs and each of those...

No, not as far as I know. I've used this behaviour for years with joins/lookups to detect if there was a match -- I'd use a non-nullable column, declare it as nullable on the input link to the join stage and, of course, in the output link and later test with IsNull() in a transform stage. The only t...

If you assigned 0 to ErrorCode in your routine, then you won't get that error message (which states that the variable hasn't been assigned).

Your routine will have 2 parameters, the first is what you pass in as a string, the second is the error code. Variables in UV-Basic are case sensitive.