Dedicated Publishing Queues

[msexton]

PDMLink 9.1 M040

We are currently running 5 CAD Workers and 2 Publishing queues (OOTB), 1 CAD Workers is located at each plant location world wide. The publish jobs are sent to the CAD Worker at the originating plant from the queue. At times this causes very long delays in publishing, especially when one of the plant locations drops 75 to 100 publish jobs in the queue at the end of their work day.

We are looking at having a customization written by a 3rd party source to have 2 dedicated queues per CAD Worker as per TAN 144658. This is the only solution recommended to us by the 3rd party.

What I would like to know is if other companies have had similar issues with publishing and what solutions were put into place to correct this?

Marshall Sexton

Pro/E and Windchill Admin
Manitowoc Cranes

[jmomber]

This sounds very similar to our setup. We had 5 cadworker servers scattered around the worldrunning two instances and they all had a processing queue - 10 workers, 10 queues. We had NOTHING BUT TROUBLE with this solution, which used the distributed cadworker functionality. Jobs continued tofail for seemingly noreasonand we would often have backups of 500-1000 jobs. We ended up dropping the distributed functionality (just recently), which increased stability (jobs fail very rarely now), but we still have quite a few jobs lining up. However, now it's only due to volume and not job failures. We have been working with PTC for months trying to get a good solution, and just recently they were able to reproduce our problem with distributed workers.

After much frustration, my recommendation when working with CADWorkers is: Stay as out of the box as you can!

[patrick.chin1]

We have 6 to 7 ProE CAD workers on 1 2X quad core 64bit Windows 2008 with 48 GIGS of ram blade. We have set up 2 more backgroud method servers to manage the separte queues of WVS1 and WVS2 groups. We have about 60 ProE users. And there is no backlog of publishing and 99% success with 1% due to ProE regeneration user issues.We used to publish 2000 to 5000 jobs per day. We toned that down due to storage space. We also have multiple foreground method severs because the WGMs pull/query from the foreground method servers.

Publishing on change and checkin with no issues of performance anymore. Our Windchill server is on RedHat 5.4 x86 64bit with 48 gigs of ram. Oracle ditto. All connected via 10GIGE backbone.

The system is designed to handle over 1000 users. If you have issues in your network being the bottleneck, I suggest terminal services for your CAD users. HP and Dell has it with fail over. All data is stored on SAN/EVA rather on hard drives. The only issue is distant cross continental transfers like North America to Asia where the WAN is not reliable.

[ptc-680841]

Happy Friday The 13th, everyone!Also, happy 2012 to you as well. If that were not enough doom, gloom and danger for one message, I am curious if companies are using dedicated publishing queues with PDMLink visualization service? If so: - How are you using it? What was the problem you were trying to solve?- What has worked well?- What has not worked well in your experience?- Do you recommend this or is itlikely to turn into a boondoggle? We are just starting to look into this and am not sure if it is a good idea or a bad idea. Thanks in advance for any info you are willing/able to provide. Andy B. PDMLink 9.1 M030 (just arrived from Ilink)Wildfire 4 (staying put for now)ProductView 9 Standard (considering moving up)