You see the backfiling working. There is nothing to prevent this*. You have to interpret it this way:
- Job 192 uses 4 cores, leaving 3 cores idling - To allow job 193 to start, it's necessary that job 192 exits as the remaining cores won't satisfy the 4 cores requirement - Job 194 uses three cores which can fit in the remaining resources I.e.: whether job 194 starts or not does not influence the start of job 193 at all. === You can try this setup with jobs requesting 4, 5 and 3 cores. The 5 slot job should block the last job requesting 3 cores in this case. === As another test, you can then lower the runtime of the 3 core job, so that it fits in the remaining time of the 4 core job. In this case backfilling should occur again, as the start of the 5 core job won't be delayed by the 3 core job (as long as the estimated runtime are estimated as best as possible). -- Reuti *) Backfilling won't happen if resource reservation is switched off. But than jobs can starve in the queue, as smaller job always may slip in. > Am 04.04.2016 um 18:59 schrieb Narsimha Reddy <[email protected]>: > > Dear Sir, > > Thank you for the reply. > > I am having total of 7 cores currently. I have created a sample job as shown > below:- > #!/bin/bash > #$ -S /bin/bash > #$ -l h_rt=01:00:00 > #$ -R y > #$ -j y > #$ -e /home/chem/test/$JOB_ID.error > #$ -o /home/chem/test/$JOB_ID.out > #$ -pe orte 4 > # print date and time > date > mpirun -np $NSLOTS hostname > > So i have created 2 sample files with 4 core each and 3 core each and the > jobs are submitted as > qsub sample4.sh; qsub sample4.sh; qsub sample2.sh > > But as per my requirement first 4 core job has to be run and the next 4 core > job has to wait till the first 4 core job gets executed as per FIFO but her > first 4 core and the 3 core job are running. Kindly help me to get FIFO based > architecture. > > My qconf ssconf output is > [locuz@ge ~]$ qconf -ssconf > algorithm default > schedule_interval 0:0:5 > maxujobs 0 > queue_sort_method load > job_load_adjustments np_load_avg=0.50 > load_adjustment_decay_time 0:7:30 > load_formula np_load_avg > schedd_job_info true > flush_submit_sec 0 > flush_finish_sec 0 > params none > reprioritize_interval 0:0:0 > halftime 168 > usage_weight_list cpu=1.000000,mem=0.000000,io=0.000000 > compensation_factor 5.000000 > weight_user 0.250000 > weight_project 0.250000 > weight_department 0.250000 > weight_job 0.250000 > weight_tickets_functional 1 > weight_tickets_share 1 > share_override_tickets TRUE > share_functional_shares TRUE > max_functional_jobs_to_schedule 200 > report_pjob_tickets TRUE > max_pending_tasks_per_job 50 > halflife_decay_list none > policy_hierarchy NONE > weight_ticket 0.010000 > weight_waiting_time 0.000000 > weight_deadline 3600000.000000 > weight_urgency 0.100000 > weight_priority 1.000000 > max_reservation 20 > default_duration 8760:00:00 > > > Also i have changed h_rt value as > h_rt 2:00:00 > > queuename qtype resv/used/tot. load_avg arch > states > --------------------------------------------------------------------------------- > [email protected] BIP 0/4/4 0.00 linux-x64 > 192 0.60500 sample4.sh chem r 04/04/2016 22:33:50 1 > 194 0.50500 sample2.sh chem r 04/04/2016 22:33:50 3 > --------------------------------------------------------------------------------- > [email protected] BIP 0/3/3 0.02 linux-x64 > 192 0.60500 sample4.sh chem r 04/04/2016 22:33:50 3 > > ############################################################################ > - PENDING JOBS - PENDING JOBS - PENDING JOBS - PENDING JOBS - PENDING JOBS > ############################################################################ > 193 0.60500 sample4.sh chem qw 04/04/2016 22:33:47 4 > > Username:- chem; queue:- all.q; nodes:- ge.test.server(3 > core),c1.test.server(4 core) > Kindly help me to resolve the issue. > > > Thanks & Regards, > A.Narsimha Reddy. > > > > Subject: Re: [SGE-discuss] Resource Reservation > > From: [email protected] > > Date: Mon, 4 Apr 2016 17:25:09 +0200 > > CC: [email protected] > > To: [email protected] > > > > > > > Am 02.04.2016 um 19:01 schrieb Narsimha Reddy <[email protected]>: > > > > > > Dear Team, > > > > > > Kindly help me out for the implementation of resource reservation in the > > > Grid Engine. > > > > > > I want this for the implementation of pure FIFO model to be applied for > > > grid engine jobs as per the job ids. > > > > Mostly you need these settings in the scheduler configuration: > > > > $ qconf -ssconf > > ... > > policy_hierarchy NONE > > ... > > max_reservation 20 > > default_duration 8760:00:00 > > > > and jobs should be submitted with `qsub -R y ...` (could be defined as > > default in "sge_request"). It's advisable to replace the default runtime > > with the real expected one. Despite the FIFO scheduling, backfilling could > > still occur as this would use resources which would idle otherwise and > > won't influence the FIFO of other jobs. > > > > -- Reuti _______________________________________________ SGE-discuss mailing list [email protected] https://arc.liv.ac.uk/mailman/listinfo/sge-discuss
