Thanks for answer, I test it again and you were right? I don’t know why that day I got two different outputs coming from Linux and Windows machines..
Thanks again for clarifying!
On Oct 28, 2011, at 5:17 PM, Jaime Frey wrote:
On Oct 28, 2011, at 9:22 AM, Alas, Alex [FEDI] wrote:
The reason I asked was when I compared the results from a Linux and a Windows machines, the output was different. The other day I troubleshoot some random jobs from one of my users, the jobs were idle for a long period of time while they were plenty idle nodes to accept the jobs. When I ran the condor_q –better-analyze, it came back with something similar to what you are posting below but when I ran condor_q –better-analyze on a Linux machine, the output was more vast and specific. That led me to check the condor_q –help and that is when I saw the –better-analyze option was gone.
condor_q -better-analyze became just condor_q -analyze in Condor 7.5.1. The -better-analyze option is still accepted and does the same thing, but is no longer displayed in the usage message.
condor_q -analyze works the same way on both unix and windows. I just tried Condor 7.6.4's condor_q on both and got the same results, pasted below.
One thing to keep in mind is that you won't get the full analysis if there aren't any machines in your pool at the moment.
We've discovered a problem with condor_q -analyze when any of the machine or job ads have an attribute value that ends in a backslash. Instead of the normal analysis of the job's requirements, you'll see the error "Unable to process machine ClassAds" or "Unable to process job ClassAd". We're working on a fix:
Thanks and regards,
UW-Madison Condor Team