Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 365330

Summary: Contributed resource manager for Lonestar (TACC)
Product: [Tools] PTP Reporter: Jeffrey Overbey <com-eclipse-dot-org>
Component: RMAssignee: Jeffrey Overbey <com-eclipse-dot-org>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: P3 CC: jalameda
Version: 6.0   
Target Milestone: ---   
Hardware: PC   
OS: Mac OS X - Carbon (unsup.)   
Whiteboard:
Bug Depends on: 365254    
Bug Blocks:    
Attachments:
Description Flags
JAXB resource manager for Lonestar g.watson: iplog-

Description Jeffrey Overbey CLA 2011-12-01 10:40:52 EST 
Created attachment 207794 [details]
JAXB resource manager for Lonestar

The attached XML file defines a resource manager for lonestar.tacc.utexas.edu.  It requires the SGE LML driver in Bug 365254.

The only part of this definition that is Lonestar-specific (besides the name) is the content in <attribute name="Resource_List.pe" type="string">.  Lonestar has 12 cores per node, and the largest queue supports a maximum of 4104 tasks, so the list of parallel environments is based these values.  A machine with a different number of cores per node and/or a different number of maximum tasks would supply a different list of PE's.
Comment 1 Greg Watson CLA 2012-02-20 10:52:41 EST 
Committed to master. I also added a generic grid engine RM that uses the same xml definition. Thanks!
Comment 2 Jeffrey Overbey CLA 2012-02-20 11:01:09 EST 
Oh, thanks, Greg.  I was going to commit this but didn't get a chance yet.

I'll take a look at the generic RM and swap out the PE list for a text box (if you haven't done that already) since that's the only Lonestar-specific part.

I also have a definition for ranger.tacc.utexas.edu that I'll commit.

Jeff
Comment 3 Jeffrey Overbey CLA 2012-02-21 15:11:36 EST 
Done.  The Generic GE RM is slightly more generic now, and I added the RM for ranger.tacc.utexas.edu.
Comment 4 Greg Watson CLA 2012-02-21 18:12:26 EST 
Cool. Didn't mean to rush you but I wanted to use it!
Comment 5 Jeffrey Overbey CLA 2012-02-21 20:58:18 EST 
No problem at all.  I definitely want people will try it on non-TACC systems before Juno; I suspect it will take a few iterations before we identify the right points of variability...