Is the bind / grant of db2esql required for DB2 z/OS?

sarahendie · Post by **sarahendie** » Wed Oct 12, 2005 8:10 pm

Can someone tell me if the bind and grant for db2esql is required for DB2 z/OS?

The Enterprise Edition installation indicates that the db2setup.sh and
db2grant.sh must be run to bind dsadm.db2esql. What I have found is that not all of the DB2 install related steps are relevent if you are using DB2 on z/OS.

The problem I'm experiencing is that I'm not able to import metadata via Orchestrate and wondered if this bind was required against my z/OS DB2 subsystem.

Eric · Post by **Eric** » Thu Oct 13, 2005 8:39 am

Are you sure that DB2 z/os is supported by the Parallel Engine? (rather than DB2 EEE)

If z/os is ok then I would expect that you would need the bind and grant as these are a requirement of the parallel DB2 stage.

track_star · Post by **track_star** » Thu Oct 13, 2005 9:04 am

The DB2 EE operator and the import utility will not connect to DB2 on any mainframe (AS/400, OS/390, z/OS). If you want to have that connectivity in a parallel job, you must use the ODBC EE operator or the DB2 API plug-in from the server canvas.

sarahendie · Post by **sarahendie** » Thu Oct 13, 2005 11:22 am

We are successfully running PX jobs on a unix AIX server to access DB2 z/OS using the DB2 API plug-in. The problem I'm experiencing is that I can't seem to get my Import , Table Definition, Orchestrate schema definitions to successfully connect to DB2 z/OS in a server project.

When I am in a USS (unix system services) project, I have no problems.

From this, I'm assuming that there is some setup on the server side that I've missed in order to get it working.

track_star · Post by **track_star** » Thu Oct 13, 2005 12:13 pm

That's what I'm trying to tell you. Orchdbutil (Orchestrate DB metadata import utility) and the db2read operator will not conect to DB2 instances on a mainframe. The DB2 API plug-in is a server-side plug-in that is wrapped for PX use, and that's why it is working for you. It can connect to mainframe DB2--it is not, however, a true parallel operator.

As for USS, it basically just runs parallel jobs on nodes of the mainframe. You still have to have PX installed on a UNIX system. You can check on your system to verify this, but I do not believe that orchdbutil gets installed to the USS system, therefore, you are using orchdbutil from your local UNIX installation (also where you are developing your jobs and compiling them). There is a section of the Installation Guide (Chapter

that details the installation and configuration necessary to get things working on a USS installation, in particular, a section that describes how to setup DB2 connectivity.

sarahendie · Post by **sarahendie** » Thu Oct 13, 2005 1:02 pm

Thanks for clarifying.

I just checked and orchdbutil is installed on our USS system so now it makes more sense why I was able to use Orchdbutil against DB2 z/OS in my USS project, but not my server project.

So, basically, I need to stop trying to get Orchdbutil to work in my server project. I'll use the DB2 plugin.

Thanks for the help.

rhys.jones@target.com · Thu Oct 13, 2005 4:49 pm

We use the DB2 API stage in a parallel canvas to connect to mainframe DB2. Underneath the covers we use DB2Connect as our middleware for DB2 API stages. Whenever a new database instance is added, the DB2Connect folks have to create/bind packages (for both mainframe and DB2 UDB on Unix). So I would imagine that if you don't use DB2Connect, then you'd need to explicitly run the create/bind package utilities.

Also, for importing metadata from the mainframe, we use the DSDB2 Plug-In. It brings up a list of all the catalogued DB2 instances (coming from DB2Connect) and we just have to specify a userid/password.