Reg:Configuration file

[ds_sai]

Hi,
I have installed Datastage 7.5x2 in Windows XP.My configuration file contents of default is:

{
node "node1"
{
fastname "Mycomputername"
pools ""
resource disk "f:/Datasets" { pools "" }
resource scratchdisk "f:/Scratch" { pools "" }
}
}

Is their anything needs to be added to the configuration file,so that all the parallel jobs will run perfectly?Could u plz help me on this regard?Thank you in Advance.


Cheers,
sai

[ray.wurlod]

Welcome aboard.

The configuration file you posted contains only one node name, so that all your jobs will run without parallelism.

To run parallel a configuration file requires more than one node name. For example:

Code: Select all

/* Two node configuration file sharing resource and scratch disks. */
/* (Best practice is to use separate file systems.)                         */
{ 
   node "node1" 
   { 
      fastname "Mycomputername" 
      pools "" 
      resource disk "f:/Datasets" { pools "" } 
      resource scratchdisk "f:/Scratch" { pools "" } 
   } 
   node "node2" 
   { 
      fastname "Mycomputername" 
      pools "" 
      resource disk "f:/Datasets" { pools "" } 
      resource scratchdisk "f:/Scratch" { pools "" } 
   } 
} 

"Run perfectly", of course, involves many other factors than the configuration file. You really need to undertake training in parallel job development or, if you are a competent server job developer, some kind of transition class. It's a totally different mind set.

[mcs@rajesh]

The basic idea behind the configuration file is to allocate the nodes to the Parallel job which makes it to run parallely and it will allocate the memory space according to the job ...
bye..

[sanath1234]

As yours is a standlone system that is default config_file your system will get so u cannot do anything about it or reconfigure. if your system is a SMP OR MMP OR cluster then u can about ot configer ie in 2or 4 node.
so for the time being thats it u can do

[quote][/quote]

[ArndW]

ds_sai,

you might be better served by ignoring sanath1234's post. The levels of parallelism you choose for PX/EE have nothing to do with a "standalone" system. Even a 1-CPU system may benefit from multiple virtual nodes. You, not the system, choose how to configure your PX/EE.

[ds_sai]

Hi all,

Thank u for u r guidance , Can anyone please tell me how to do in my system
of Configuration file,mean by giving the multiple nodes?So that i too under the parallelism concept.Thank you very much in Advance.


Cheers
Sai

[ds_sai]

Hi all,

Thank u for u r guidance , Can anyone please tell me how to do in my system
of Configuration file,mean by giving the multiple nodes?So that i too under the parallelism concept.Thank you very much in Advance.


Cheers
Sai

[ArndW]

Since Ray's example might not be visible to you, I suggest you open up the PDF Parallel Job Developer's Guide and read Chapter "The Parallel Engine
Configuration File", perhaps looking at one of the sample multinode files such as the one on page 58-24.

[aakashahuja]

If you want to visualise (not the right word but can give you the context) parallelism then you can try the following along with what ArndW suggested:

First, run any sample job with a one node configuration. Watch the job monitor.

Then, run with a 2 node config and then watch the monitor. You will see multiple instances of stages (set to run in parallel in your job), equal to the number of nodes (assuming that you have not used constraints).

Hope this helps...

Cheers
Aakash

[aakashahuja]

If you want to visualise (not the right word but can give you the context) parallelism then you can try the following along with what ArndW suggested:

First, run any sample job with a one node configuration. Watch the job monitor.

Then, run with a 2 node config and then watch the monitor. You will see multiple instances of stages (set to run in parallel in your job), equal to the number of nodes (assuming that you have not used constraints).

Hope this helps...

Cheers
Aakash

[sanath1234]

ds_sai,

you might be better served by ignoring sanath1234's post. The levels of parallelism you choose for PX/EE have nothing to do with a "standalone" system. Even a 1-CPU system may benefit from ...

"The Parallel Engine
Configuration File" in the developers guide will defenetly help u but problem is the OS u are using is XP i tryed but failed and tryed it on 2003 server i can create a 2 node config file but i am unable to use the file.
any comments on this will be appreciated

[ArndW]

sanath1234 - explain

i tryed but failed

- what didn't work?

[sanath1234]

sanath1234 - explain

i tryed but failed

- what didn't work? ...

i configured in 2node on my system which is on windows 2003 server by reading the help but when i tryed to use this new config file in stages it is disabled

do i have to do anything like adding any environmentel variables at project level or anything else (rest working fine)
thnx in adv..

[balajisr]

You need to set APT_CONFIG_FILE environment variable to the new config file in your job or at project level.

[ds_sai]

Hi all,
Thank you very much for one and all,for advising me about the parallelism.

I got one doult about data stage configuring the nodes.I am having dual o.s in my system,i.e 1.O.S:Windows XP professional 2.O.S:Windows Vista business.Based on this is it possible to create multiple nodes in my system?
If yes means can anyone please guide me with your valuable ideas how to proceed?


Cheers,
ds