DSXchange

I have two configurations:

node1:

{
node "node1"
{
fastname "tnet368"
pools ""
resource disk "/prop/loc/tnbilfor/dtstnbilfor/work/node01/disk00/" {pools ""}
resource scratchdisk "/prop/loc/tnbilfor/dtstnbilfor/work/node01/scratch00/" {pools ""}
}

}


node2:

{
node "node2"
{
fastname "tnet368"
pools ""
resource disk "/prop/loc/tnbilfor/dtstnbilfor/work/node02/disk00/" {pools ""}
resource scratchdisk "/prop/loc/tnbilfor/dtstnbilfor/work/node02/scratch00/" {pools ""}
}

}


If I run a parallel job with the first configuration, it runs fine. if I run the same job with the seconde configuration, it aborts with:

descriptions: Error when checking operator: Node name "node1" not in config file

If I reset the job and run it again with the second, it runs fine. If I run then with the first, it aborts.

It seems that the engine remembers the last configuration and excepts that the same node is available.

Can anybody help?

I couldn't reproduce that here. Which "operator" is giving the error?

hne wrote:I have two configurations:

node1:

{
node "node1"
{
fastname "tnet368"
pools ""
resource disk "/prop/loc/tnbilfor/dtstnbilfor/work/node01/disk00/" {pools ""}
resource scratchdisk "/prop/loc/tnbilfor/dtstnbilfor/work/node01/scratch00/" {pools ""}
}

}


node2:

{
node "node2"
{
fastname "tnet368"
pools ""
resource disk "/prop/loc/tnbilfor/dtstnbilfor/work/node02/disk00/" {pools ""}
resource scratchdisk "/prop/loc/tnbilfor/dtstnbilfor/work/node02/scratch00/" {pools ""}
}

}


If I run a parallel job with the first configuration, it runs fine. if I run the same job with the seconde configuration, it aborts with:

descriptions: Error when checking operator: Node name "node1" not in config file

If I reset the job and run it again with the second, it runs fine. If I run then with the first, it aborts.

It seems that the engine remembers the last configuration and excepts that the same node is available.

Can anybody help?

Hi ,

Since configuration is related to server . You need to restart the server for the configuartion file to get updated.

Regards
Thiru

samythiru - no, there is no need to restart the DataStage engine when using different PX configuration files.

p.s. No need to quote messages, either.

Is your job design using Data Set or File Set stage(s)? These are, in some sense, tied to the configuration file that was used when they were created. In particular, the resource disk that specifies the location of their data files must be accessible in whatever configuration is used subsequent to their creation to access them. This is not the case between your configuration files.

This will happen if your job has a dataset or a file set stage and

1. You have used the 'overwrite' option.
2. The data set or file set already exists.

So basically then PX tries to find the part file (cant remember the correct term for the data file on a particular partiiton) to delete, it will not be able to find it since a node name has changed. Then it throws the error.

Cheers
Aakash