DSXchange

Sorry, I thought I had made it an un-premium post as it didn't really answer your question. I've changed it now.

Juls, the dsjob command lets you specify server, userid and password to do any of the actions on a remote machine. Please look at page 19-140 of the Server Job Developer's Guide for more information.

I think you need to stop using a microsoft outlook system for your mail. The file is not really sent but the way the MS mail server works with smtp makes this empty attachment get appended to the message. I know that you can turn this off but can't recall how - check MS pages for a description on ho...

runtime column propagation looks like the probable culprit.

The commit might affect your results if you are reading a record and then writing it back or if you are doing multiple updates to the same record. This error shouldn't be occurring unless you do have one of the two aforementioned conditions - or have another process working on the table as well.

Option #2 won't commit until the end of the job.

What exact error are you getting with #1? Are you reading these records as well? Which database is behind the ODBC layer and are you doing updates, upserts or inserts?

You can also do a TRIM(In.Column,'0','L') which will remove all leading "0" characters from your column.

Can you modify a copy of your job so that it doesn nothing apart from read your data and write it to /dev/null? That will give you the maximum speed and you can start narrowing down your possible sources of slowdowns. Also, are you actually reading your 28-row reference table for each data row and n...

I wouldn't circumvent the advantages of chartered membership by sending the rest via e-mail. But in this case there wasn't anything much after that. You will need to put your validation logic into a job sequence and modify/fix the parameter values and then pass that (validated) value to your actual ...

You are creating a disk snapshot of the project, not of the jobs - so the state would be "running" but there would actually be no running processes after a restore and you would need to reset that. The lock ensures that the physical hashed files are in a consistant state, but doesn't do anything abo...

If I understand the problem you are modifying a job's parameter from inside the job itself (in the job control). This is not allowed and would explain why your changes aren't "seen" in other parts of the job. You will need to do your validation in a sequence and then pass the corrected values to the...

When writing to a hashed file you can increase speed by specifying a larger MINIMUM.MODULUS for the dynamic file, this can more than double the write speed. If your hashed file does not fit into your defined memory limits it will not load - you will see a message in the log file that tells you as mu...

The times depend upon repartitioning after the read as well, and we don't know what has been done with the data.

The DataStage engine has a facility to freeze all activity for a period of time, so if your backup mechanism is fast (or fast enough) you can activate this from dsadm/root with "uv -admin -lock" to immediately suspend all hashed I/O activity and "uv -admin -unlock" to restart it. The processes just ...

Never use anything less than a couple of minutes runtime for any timing comparison purposes! Your startup-time accounts for almost all your activity, so of course using more readers is going to add significantly to your times. Don't even think about using multiple processes before your runtime goes ...