:) That’s absolutely true… but secondary. The SLURM job won’t end itself, but 
it has been chewing CPU in any case. There is no runaway process on the compute 
node — it’s clean and clear.

Does anyone else ever see this?

Cheers,

Alden

> On Jun 9, 2017, at 12:17 PM, Michael Jennings <[email protected]> wrote:
> 
> 
> On Thursday, 08 June 2017, at 11:38:20 (-0700),
> Stradling, Alden Reid (ars9ac) wrote:
> 
>>   JobName      State               Start    Elapsed    CPUTime 
>> ---------- ---------- ------------------- ---------- ----------
>> PBMC_6c_0+  SUSPENDED 2017-06-06T16:39:51 1-21:12:55 15-01:43:20 
>> 
>> That 15 days??? not really possible since the job started two days ago.
> 
> Yes, it is actually possible.  Notice that you're looking at CPU TIME,
> not Elapsed Time (the prior column).  Your job used 8 CPUs, so you
> have to multiply the elapsed time (just over 45 hours) by the number
> of CPUs you used.  If you do that, you'll get 1,302,072 CPU seconds,
> or just over 15 days. :-)
> 
> Michael
> 
> -- 
> Michael E. Jennings <[email protected]>
> HPC Systems Team, Los Alamos National Laboratory
> Bldg. 03-2327, Rm. 2341     W: +1 (505) 606-0605

Reply via email to