[torqueusers] Processors marked as free while not

Garrick Staples garrick at usc.edu
Wed Feb 23 09:51:40 MST 2005


On Wed, Feb 23, 2005 at 10:03:35AM +0100, tegner alleged:
> We are running maui-3.2.6p9 and torque-1.1.0p4, and at some point in 
> time the system (torque) got confused, and somehow believies that there 
> are more free nodes than what is actually the case.

You mean that the load average on a node is higher than $max_load but
'pbsnodes -a' lists the node state as "free" instead of "busy"?

 
> Two questions:
> 
> 1. What could be the reason for this?

If it's the problem above, then it's a bug that is fixed in 1.2.0p1.  


> 2. Is it possible to "reset" the system without bringing down the 
> running jobs?

You can use 'qmgr' to set the node state to whatever you feel is appropriate.


-- 
Garrick Staples, Linux/HPCC Administrator
University of Southern California
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://www.supercluster.org/pipermail/torqueusers/attachments/20050223/26cfb9a1/attachment.bin


More information about the torqueusers mailing list