Interesting observation of Datastage Version 8

abc123 · Post by **abc123** » Mon Oct 08, 2007 3:27 pm

Writing to sequential files as target causes a "No conductor node found in the export pool" error. The way to get around this is to make all sequential file targets as datasets and the problem goes away.

lstsaur · Post by **lstsaur** » Mon Oct 08, 2007 4:38 pm

That's not correct answer. I could tell that you are running your jobs in a GRID environment since you got "No conductor node found in the export pool" error. You have to specify the path in your $APT_GRID_SEQFILE_HOST.

abc123 · Post by **abc123** » Mon Oct 08, 2007 6:41 pm

I am assuming that, by path, you mean the path where the sequential file is going to reside. I specified that path. I still get the same error.

ray.wurlod · Post by **ray.wurlod** » Mon Oct 08, 2007 8:54 pm

It's not an error, it's an alert message.

If you define a node pool called "export", whether or not you are in a grid environment, any export operator (for example a Sequential File stage that has an input link) will execute in that node pool.

You are being alerted to the fact that your configuration file does not contain a node pool called "export" or that, if it does, it does not include the conductor node.

abc123 · Post by **abc123** » Tue Oct 09, 2007 9:47 am

Ray, I get a fatal error for each sequential file stage and the job aborts. The target sequential files are not produced.

ray.wurlod · Post by **ray.wurlod** » Tue Oct 09, 2007 2:40 pm

Test or post your configuration file.

If you have no export node pool DataStage will use the default node pool. That particular alert message is non fatal.

There is something else wrong - maybe that you lack write permission where you're trying to write the output files.

abc123 · Post by **abc123** » Tue Oct 09, 2007 3:45 pm

I am in a Linux grid environment and the config file is created dynamically by an application developed by IBM. The script is called Dynamic_grid.sh.

lstsaur · Post by **lstsaur** » Tue Oct 09, 2007 5:16 pm

As I told you in earlier note, put your $APT_GRID_SEQFILE_HOST parameter in your job's sequential stage (target). Did your project lead, Jeff, tell you which directory to put your sequential file?

abc123 · Post by **abc123** » Tue Oct 09, 2007 9:21 pm

No Leo. Is there a specific directory that you need to write all sequential files in? In your earlier post, you mentioned that the value of the parameter needs to be the path where the target sequential file needs to be written. I put in the path but I still get the same error.

According to the description of the parameter when you add it in the Job Parameters section, this parameter's value should be "First Host name identified by Grid engine or from IONODE names (job parameter returned)".

Is there any documentation that you got from IBM that you can share with us about this?

srekant · Post by **srekant** » Wed Oct 10, 2007 1:21 pm

abc123 wrote:No Leo. Is there a specific directory that you need to write all sequential files in? In your earlier post, you mentioned that the value of the parameter needs to be the path where the target sequential file needs to be written. I put in the path but I still get the same error.

According to the description of the parameter when you add it in the Job Parameters section, this parameter's value should be "First Host name identified by Grid engine or from IONODE names (job parameter returned)".

Is there any documentation that you got from IBM that you can share with us about this?

Hi you need to just add the GRID parameter($APT_SEQFILE_HOST) to your job and then prefis it to you sequential file path in the sequential file stage.

The value for this parameter is the 1st value outputted by the execution of GRID_SEQUENCER.sh script.

Ex:
$APT_SEQFILE_HOST=Field(commandstage name.$CommandOutput," ",1)