[GE users] scheduler runs performance sge6.0u6 vs. sge6.0u7

Reuti reuti at staff.uni-marburg.de
Mon Dec 12 19:43:29 GMT 2005


Hi,

this is a really short schedule_interval. Did you try to set it to a  
minute or so and adjust the value of flush_submit_sec to one or two  
seconds instead (seems you are looking for an immediate scheduling)?

-- Reuti


Am 12.12.2005 um 19:34 schrieb Nicolas Joly:

>
> Hi,
>
> We are running a small cluster (14 nodes) of x86_64 machines running
> CentOS 4.2 (RHEL clone). On this cluster, with SGE6.0u6, we have setup
> a single cluster queue for interactive jobs only ... The
> schedule_interval parameter is currently set to 00:00:01.
>
> This morning, we've updated to SGE6.0u7 without changing anything in
> the configuration, and noticed that sched runs were much longer than
> previously (3.12s >>> 0.12ss).
>
> I didn't noticed anything wrong except that all the jobs are now
> staying in queue for about 3 second ...
>
> Here follow some profilling samples got from both version :
>
> 1) SGE6.0u6 (no delay)
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: static urgency  
> took 0.000 s
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: job ticket  
> calculation: init: 0.000 s, pass 0: 0.000 s, pass 1: 0.000, pass2:  
> 0.000, calc: 0.000 s
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: job ticket  
> calculation: init: 0.000 s, pass 0: 0.000 s, pass 1: 0.000, pass2:  
> 0.000, calc: 0.000 s
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: normalizing job  
> tickets took 0.000 s
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: create active job  
> orders: 0.010 s
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: job-order  
> calculation took 0.010 s
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: job sorting took  
> 0.000 s
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: job dispatching  
> took 0.010 s
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: create pending job  
> orders: 0.000 s
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: scheduled in 0.080  
> (u 0.000 + s 0.000 = 0.000): 2 sequential, 0 parallel, 1 orders, 15  
> H, 10 Q, 13 QA, 0 J(qw), 10 J(r), 0 J(s), 0 J(h), 0 J(e), 3 J(x),  
> 13 J(all), 46 C, 1 ACL, 1 PE, 2 U, 1 D, 0 PRJ, 0 ST, 0 CKPT, 0 RU,  
> 3 gMes, 0 jMes
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: send orders and  
> cleanup took: 0.040 (u 0.000,s 0.000) s
> 12/12/2005 19:04:08|schedd|raclette-adm1|I|PROF: schedd run took:  
> 0.120 s (init: 0.000 s, copy: 0.000 s, run:0.120, free: 0.000 s,  
> jobs: 13, categories: 1/1)
> 12/12/2005 19:04:09|schedd|raclette-adm1|I|PROF: sge_mirror  
> processed 35 events in 0.000 s
>
> 2) SGE6.0u7 (3 sec delay)
> 12/12/2005 16:41:23|schedd|raclette-adm1|P|PROF: static urgency  
> took 0.000 s
> 12/12/2005 16:41:23|schedd|raclette-adm1|P|PROF: job ticket  
> calculation: init: 0.000 s, pass 0: 0.000 s, pass 1: 0.000, pass2:  
> 0.000, calc: 0.000 s
> 12/12/2005 16:41:23|schedd|raclette-adm1|P|PROF: job ticket  
> calculation: init: 0.000 s, pass 0: 0.000 s, pass 1: 0.000, pass2:  
> 0.000, calc: 0.000 s
> 12/12/2005 16:41:23|schedd|raclette-adm1|P|PROF: normalizing job  
> tickets took 0.000 s
> 12/12/2005 16:41:23|schedd|raclette-adm1|P|PROF: create active job  
> orders: 0.000 s
> 12/12/2005 16:41:23|schedd|raclette-adm1|P|PROF: job-order  
> calculation took 0.010 s
> 12/12/2005 16:41:23|schedd|raclette-adm1|P|PROF: job sorting took  
> 0.000 s
> 12/12/2005 16:41:23|schedd|raclette-adm1|P|PROF: job dispatching  
> took 0.000 s (2 fast, 0 comp, 0 pe, 0 res)
> 12/12/2005 16:41:23|schedd|raclette-adm1|P|PROF: create pending job  
> orders: 0.000 s
> 12/12/2005 16:41:23|schedd|raclette-adm1|P|PROF: scheduled in 0.030  
> (u 0.000 + s 0.000 = 0.000): 2 sequential, 0 parallel, 11 orders,  
> 15 H, 13 Q, 13 QA, 0 J(qw), 6 J(r), 0 J(s), 0 J(h), 0 J(e), 2 J(x),  
> 8 J(all), 46 C, 1 ACL, 1 PE, 2 U, 1 D, 0 PRJ, 0 ST, 0 CKPT, 0 RU, 1  
> gMes, 0 jMes, 0/0 pre-send, 0/0/0 pe-alg
>
> 12/12/2005 16:41:26|schedd|raclette-adm1|P|PROF: send orders and  
> cleanup took: 3.090 (u 0.010,s 0.000) s
> 12/12/2005 16:41:26|schedd|raclette-adm1|P|PROF: schedd run took:  
> 3.120 s (init: 0.000 s, copy: 0.000 s, run:3.120, free: 0.000 s,  
> jobs: 8, categories: 1/1)
> 12/12/2005 16:41:26|schedd|raclette-adm1|P|PROF: sge_mirror  
> processed 63 events in 0.000 s
>
>
> Thanks in advance,
> Regards.
>
> -- 
> Nicolas Joly
>
> Biological Software and Databanks.
> Institut Pasteur, Paris.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list