DSXchange

The Configuration you mentioned is compatable for 7.5x.

I know the APT_CONFIG_FILE format is compatible.

However, if the server name changed, we simply transfer the disks to new server, the Dataset can be read correctly ?

Yes, you can move without worries in this part of the transition. The dataset descriptors don't contain a a server name reference, just the schema information and the paths to the actual data files.

I'm confused by ArndW's reply. I see the config file in a dataset when I do a orchadmin ll. The fastname will be the old server. If the job that generates the dataset does NOT use "Use Existing (Discard schema and records)" when the job is run, then won't the existing dataset attempt to use the old environment's config file and not the new?

HI,

The config file information is from 'Orchadmin' command and not from dataset descriptor file.

Hey, I'm still confused. Maybe you can help me out. I see this doing a orchadmin ll:
"Embedded ORCHESTRATE system configuration:"

and then the config information.

so my questions would be these;

1. How does orchadmin know which configuration file applies to the dataset? We have several and I would consider that quite mis-leading if orchadmin showed me one when in fact the dataset was created with another.
2. Why does orchadmin say "embedded"?

Thanks for the help!

Is this a hijack of the thread? Maybe, but the answer will apply.

The configuration file used/reported is whatever is currently pointed to by the APT_CONFIG_FILE environment variable.

Still confused. I was content with Ray's answer as that made sense and he is Ray after all. However, perusing the documentation for another problem I took a quick peek at Managing DataSets from the PX Developer's Guide pdf page 1046. It very clearly states that the configuration file in (sic) use when the DataSet was created is in the Descriptor file.

Having it there makes sense to me since the Config file in the current APT_CONFIG_FILE env variable could be radically different then the config file used when the dataset was created.

I took a look at the files and trobinson is correct, the config file that was used is embedded in the dataset along with the schema and file locations.
But looking at the orchadmin options it seems that the locations listed as the paths to the data files override any config file inherited attributes, so you will be able to utilize the dataset on the new node.

You could test this by changing the config settings in the dataset to some other path and see if the datasets is still displayable.

I will surmise that you go on to discuss the -x option of orchadmin. Doing orchadmin -help discusses which config file to use when using orchadmin; the one contained within the dataset descriptor file or the APT_CONFIG_FILE one.

Reading further it looks like whatever one does in DataStage; NEVER, ever ...

Clearly an embedded configuration is necessary in a Data Set or a File Set. For example, if it was written under control of a four-node configuration file, then it can only be read from those four nodes, irrespective of the "current" configuration file. This can imply a need to re-partition in a running job.

My statement was about configuration file - I should have been more explicit.

Yes, clearly... now. Remember the single most important success factor in any DataStage implementation is ...












... the use of the ellipse.