Spectrum LSF

Showing 502

Create a way to prevent a job with an SLA from using another SLA's hosts/slots/memory if the other SLA is under-committed

LSF hosts come and go, and when you create an allocation that uses exact numbers, there are times when a Pool can be over and under committed. In the case of an under-committed pool today, consumers from other pools can leverage free slots from th...

over 1 year ago in Spectrum LSF / Administration 0 Future consideration

2 VOTE

Add Effective Memory Utilization to All Statistics Tables

RTM maintains a consistent set of ETL's across multiple dimensions. It is the same table structure for all of these ETL's. However, a key metric is missing from the ETL's and it's the effective memory utilization. In other words sum(max_memory) /s...

about 3 years ago in Spectrum LSF / Operations/RTM 1 Not under consideration

2 VOTE

When a Benchmark job fails It would be nice to see the bsub used to submit the job

We can manually construct the actual bsub, but it would be nice to be able to cut and paste it when editing the Benchmark job so we can test that it actually will work. In one case we were getting "Queue Permission Errors" and it having that bsub ...

about 3 years ago in Spectrum LSF / Reporting/Explorer 2 Not under consideration

1 VOTE

Allow RTM to show jobs 'submitted' between ranges from job details page

We are in the middle of a bench-marking activity and it occurred to me that there is now way to say in RTM show me jobs that were submitted between these two dates. It would be helpful to have that filter available.

over 1 year ago in Spectrum LSF / Operations/RTM 1 Not under consideration

1 VOTE

Ability to easily load missing LSF event streams and LSF accounting data

Customer has gaps in data loading due to problems with ElasticSearch and is looking to fill these gaps with some way of easily loading the missing data to Elastic Search. One idea would be to add a command to load data in a specified date and time...

over 1 year ago in Spectrum LSF / Reporting/Explorer 1 Not under consideration

1 VOTE

bacct command output should have GPU memory Usage details visible like CPU MEM usage

currently in bacct command output we could see below: Accounting information about this job: Share group charged </hkorva> CPU_T WAIT TURNAROUND STATUS HOG_FACTOR MEM SWAP1844603.00 0 73687 done 25.0329 171.1G 2M if the job is a GPU job then...

over 1 year ago in Spectrum LSF / Administration 0 Future consideration

1 VOTE

Allow Session Scheduler to support the queue level Job Starter when the Session Spreads to Multiple Hosts

In testing of Session Scheduler, it works perfectly, on secondary hosts, it does not seem to respect our job starter which will limit our ability to deploy Session Scheduler more broadly. Closing this gap is important.

over 1 year ago in Spectrum LSF / Administration 0 Future consideration

1 VOTE

Allows Session Scheduler Account Directory and moving parts to support the lsf.conf variable LSB_STDOUT_DIRECT.

We recently were able to confirm that if the user leaves off the SSCHED_ACCT_DIR in lsb.params, that they can set the same as an environment variable, and eventually the Session Scheduler account file will end up in that location, but the users re...

over 1 year ago in Spectrum LSF / Administration 1 Not under consideration

1 VOTE

Add user group filter for bacct and bhist

To evaluate jobs w.r.t. user groups/fair share groups I would like to have a filter for user groups in bacct and bhist similar to the option -G of bjobs.

over 1 year ago in Spectrum LSF / Reporting/Explorer 0 Future consideration

4 VOTE

PSUSP jobs need JOB_CONTROLS

Users need notifications when jobs are PSUSP due to pre-exec failures, and JOB_CONTROLS do not work because the job isn't RUN when it is suspended. We also need a tunable parameter to have these jobs killed after some amount of time.

over 6 years ago in Spectrum LSF 2 Not under consideration

Please enter your email address

FILTER BY CATEGORY