Hi, Dave,
I tested only mem_free in my environment. We're on el6.3 running SGE
8.1.1, x86_64. My symptoms came from jobs not dispatching and turning
schedd_jobinfo on to reveal that that mem_free was an unknown parameter
on my hosts. All I know is after adding the attribute to the
complex_values list for my exec hosts, we successfully worked-around the
issue. I'm working on getting a virtualized development environment
online to test against. Once I do, I'll provide more output on the problem.
-Brian
Brian Smith
Assistant Director, Research Computing
Information Technology
University of South Florida
4202 E. Fowler Ave. SVC4010
Office Phone: +1 813 974-1467
Organization URL: http://rc.usf.edu
On 09/16/2012 08:44 AM, Dave Love wrote:
Brian Smith <[email protected]> writes:
Hi, Joseph,
Wow. Just ran into this one myself (mem_free). Adding a complex value to
each host set to mem_free=(ram on host) made for a good work-around. The
variable is still being reported, so it still reports the real mem_free
(not the value you set), but the problem seems to go away when I do this.
Just FYI.
Is the report you get from qalter -w p the same as previously reported?
Is this specific to mem_free or more general, and can anyone reproduce
it in a basically default configuration? It's difficult to address when
it's not reproducible, and doesn't seem worth spending much time on.
_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users