[torqueusers] RFC: job-state diagram

Caird, Andrew J acaird at umich.edu
Tue Jun 19 09:28:41 MDT 2007


> -----Original Message-----
> From: torqueusers-bounces at supercluster.org 
> [mailto:torqueusers-bounces at supercluster.org] On Behalf Of Paul Millar
> Sent: Tuesday, June 19, 2007 11:25 AM
> To: torqueusers at supercluster.org
> Subject: Re: [torqueusers] RFC: job-state diagram
> 
> On Monday 18 June 2007 21:35, Garrick Staples wrote:
> > On Mon, Jun 18, 2007 at 05:57:56PM +0100, Paul Millar alleged:
> > > I've been trying to understand Torque and, in particular, both the
> > > semantics of the different job states and under what 
> conditions a job
> > > will change state.
> >
> > Note that job states are defined in IEEE 1003.1,
> >
> > http://www.unix.org/single_unix_specification/
> >
> > Shell & Interfaces -> Batch Environment Services
> 
> Thanks for the information, it was very useful.
> 
> There are some details that 1003.1 leaves as 
> implementation-specific; for 
> example, job states when in routing queue.  Also, Torque 
> deviates from IEEE 
> 1003.1 by including an extra state "suspended".

I believe Torque also has a "C" state for "completed" - I didn't notice
that in the spec.

--andy


More information about the torqueusers mailing list