Skip to main content
B
byork17-Peridot
17-Peridot
January 14, 2016
Solved

Multiple APE Servers for PDMlink

  • January 14, 2016
  • 1 reply
  • 3484 views

I was wondering if there are other ways to configure PE to publish multiple jobs from PDMlink other than setting up another PE server.  I was wondering if there would be a way to do that with sub processes?

Any help would be greatly appreciated! 

- Brian

Best answer by SirkoRudolph

Brian,

in PDMLink Site administration UI there is a link for Worker Agent Administration.

You need to click the "configure" link in that toll and click the "Edit" button.

Select your APE worker from the list of configured workers.

Clicking through a few pages you get to the page where you see a textbox "Worker Instances". Just enter the appropriate number here, save and reload the agent configuration file.

What I do not remember from the top of my head though is how PDMLink / APE publishing behaves as to when jobs are removed from the queue. You might also need to add Publising queues for each worker instance you add to get real parallel processing.

You should use Windchill HelpCenter to lookup how to add Publishing Queues.

Hope that helps.


Sirko

1 reply

C
ClayHelberg18-Opal
18-Opal
January 14, 2016

Hi Brian--

Yes, that's just what subprocesses are for. You can configure this in the e3config.xml file. Look at the comments in that file, they include lots of useful information on configuring subprocess pools.

Note that you will want to check the hardware configuration (virtual or real) on your PE server to make sure it has what it needs to support what you want. Ideally, you would match the number of CPU cores to the number of subprocesses you need to support. If you need to be able to process 4 jobs simultaneously, make sure the PE server has 4 cores available to PE. Any less, and you will add extra overhead for task switching. (Any more doesn't hurt, but doesn't help either, meaning you're wasting resources.) You'll want to check memory requirements as well, though those can be harder to pin down, since they depend a lot on the types of jobs (documents) you're processing.

--Clay

B
byork17-PeridotAuthor
17-Peridot
January 14, 2016

Clay,

Thanks for the info! I have another question around the sub process.  Do I need to setup separate queues in PDMLink or just new workers specifying the different sub processes?

Thanks,

Brian

C
ClayHelberg18-Opal
18-Opal
January 14, 2016

Hi Brian--

I'm not sure about the PDMLink configuration. I guess it depends whether you want PDMLink to manage the queuing, or if you want PE to do it. If you don't do any queuing on the client side (PDMLink in this case), you can just throw jobs at PE as they come up, and PE will manage the queuing, processing the first N jobs concurrently (where N is the number of subprocesses you've configured for the pool), and queue the rest for subsequent processing. PE handles all subprocess management internally, so I don't know of a way to make PDMLink know or care about the PE subprocesses. I don't think you can make PDMLink workers "specify different subprocesses" in PE, and even if you could, I'm not sure I see a reason to.

If you really need for PDMLink to be able to keep composition jobs separate, you might be able to do something like that by setting up a distinct subprocess pool in PE for each PDMLink queue.

--Clay

Terms & ConditionsAccessibility statement