Spectrum LSF

Showing 501 of 15368

For more accurate pend time

Product: IBM Spectrum Computing AnalyticsVersion: 9.1.4Actor: SKhynix, Korea HQ Customer want to know how many server's slot was not used and application license was pended and what user could not run own jobs in how long time. To find some inform...

almost 7 years ago in Spectrum LSF 1 Not under consideration

2 VOTE

LSF should support mapping resources to host group(s)

We need to map 200 dynamic gpu resources to all 120 gpu nodes due to gpu auto config is not supported in RHEL6. In the current latest LSF 10.1.0.9, we have to add 120 host names to each of 200 resources. This makes lsf.cluster file very messy. And...

over 4 years ago in Spectrum LSF / Administration 0 Not under consideration

1 VOTE

allow update for span string after a job is started

Hello, currently, it's not possible to update the span string after a job is started. for ex: a user start a multinodes jobs using bsub .... -n x -R ".. span[ptile=y]" ... unfortunately, he forgot to specify to the tool to start in mutlinodes an...

over 2 years ago in Spectrum LSF / Scheduling 1 Not under consideration

4 VOTE

LSF to preempt tasks in a running parallel job that use borrowed resources to allow PEND jobs with guarantees to go running

We hare 3 queues.All queues have same priority.Each queue has a GRP SLA (guarantee).All queues are setup such that each queue loans out it guarantees to the other 2 queues when not used. In customer environment is often the case that 1 queue goes ...

over 9 years ago in Spectrum LSF 2 Not under consideration

3 VOTE

Change the name of "libevent.so".

We have "libevent.so" for our application. LSF "libevent.so" conflict with it and we have to set library path for ours in job scripts. "libevent.so" is too common name, so please change it in LSF, like "liblsfevent.so".

about 7 years ago in Spectrum LSF 2 Not under consideration

1 VOTE

Preemption support for MIG GPU jobs

We would like that preemption will be supported also for MIG jobs(A100 dgx mig devices). Currently the preemption is not working for this type of jobs. Example scenatiois: kill two or more jobs and reconfigure mig to have the new job run. We need ...

over 2 years ago in Spectrum LSF / Scheduling 2 Future consideration

1 VOTE

Allow bjobs command to show the pre and post exec exitStatus/exitCode information

Right now, we require bhist to get any detail on pre/post exec error codes, it would be nice if we could get this from the job record via bjobs vs. bhist. bhist in very large environments is a non-starter for obtaining this status information.

over 2 years ago in Spectrum LSF / Administration 1 Not under consideration

4 VOTE

Roll back job submission when bsub cannot receive the jobid from mbatchd

Due to a network hiccup caused by an overloaded server, job submission using bsub returned this message which of course does not include the new jobid: Failed in an LSF library call: Communication time out. Job not submitted However the job was re...

over 9 years ago in Spectrum LSF 2 Not under consideration

1 VOTE

Allow Users to specify a ClearCase 'cleartool' command but require that it be white listed first

Allow the users to override the CLEARTOOL binary path, but only if it's included in a white list configuration file in the $LSF_ENVDIR location. I would suggest something like: cat lsf.daemons.wrap.conf /usr/altira/bin/cleartool /some/authorized/...

over 2 years ago in Spectrum LSF / Administration 1 Not under consideration

1 VOTE

Set more job variables in parent SBATCHD prior to forking the job SBATCHD/RES

I've been working with the Daemons Wrap source code trying to make the logging more presentable, and I've come accross a hurdle to effectively enhancing that integration. I need both the Job Start and Submit Times, and normally, I would just be ab...

over 2 years ago in Spectrum LSF / Administration 1 Not under consideration

Please enter your email address

FILTER BY CATEGORY