High-Memory Queues
Details on Jobs that Produced a Warning
As of Mon Nov 17 09:07:06 EST 2025
jobID.tID jobName user queue - mem_res max(vmem) - vmem avgmem cpu age eff. nslots
------------------------------------------------------------------------------------------------------------------------
11042543.1 xadm uribeje lThM.q - 400.0G >> 32.483G - 23.810G 19.096G 4.4d 4.4d 5.0% 20/mthread
11043155.1 Delphinidae_IQT mcgowenm lThM.q - 900.0G >> 23.109G - 10.756G 8.909G 3.0d 3.0d 16.6% 6/mthread
11043499.1 xvcf2 uribeje uThM.q - 400.0G >> 28.062G - 26.230G 18.464G 2.0d 20.4h 29.2% 8/mthread
11043498.1 xvcf2 uribeje uThM.q - 400.0G >> 27.834G - 26.083G 18.256G 2.0d 20.4h 28.8% 8/mthread
11043497.1 xvcf2 uribeje uThM.q - 400.0G >> 28.473G - 28.473G 18.741G 2.0d 20.4h 29.1% 8/mthread
11043501.1 xvcf2 uribeje uThM.q - 400.0G >> 28.429G - 26.290G 18.588G 2.0d 20.4h 29.3% 8/mthread
|
A warning is generated if either:
• too much or too little memory is reserved:
mem_res versus max(vmem); or
• the job efficiency is too low or is too high.
Click on the link under the jobID.tID
heading to view the job's corresponding graph.
The quantity
mem_res is the amount of memory reserved for the job, while
max(vmem) is the maximum amount of memory a job has used (so far);
to optimize the cluster's memory usage, these two numbers should be similar.
The job efficiency, eff., is the amount of CPU used so far divided by the product of the
age by the number of slots.
• A low efficiency means that the job is not using all the allocated CPUs (slots);
• a value above 100% means that the job is using more CPUs cycles
(threads) than the requested number of slots (nslots).
|
This page was last updated on
Monday, 17-Nov-2025 09:11:52 EST
with mk-web-page.pl ver. 7.3/1 (Oct 2025/SGK)
|