Scheduling

Showing 26

Allow preemption of shared gpu jobs (j_exclusive=no) by higher priority job queues.

We need the LSF support for preemption of shared gpu jobs. Currently GPU preemption is supported only for jobs in mode j_exclusive=yes This is a very big problem as we have a significant waste of gpu memory resource. because if a lower priority jo...

about 1 month ago in Spectrum LSF / Scheduling 1 Not under consideration

2 VOTE

Enable option for hardcoding GPU memory request like its available for CPU memory request

At this moment IBM offers features of hardcoding CPU memory. We have implemented this feature in our cluster with esub scripting. Similarly we want to have same feature available for GPU memmory (gmem) to be hardcoded. I have raised ticket to our ...

about 2 months ago in Spectrum LSF / Scheduling 2 Not under consideration

1 VOTE

Listening to jobstates

We submit our jobs to LSF with the LSF API (C++, lsbatch.h). Our software has to give a feedback to the users about the current state of the jobs. For this query we call lsb_readjobinfo for each active job in our list. It can take a long time to c...

4 months ago in Spectrum LSF / Scheduling 0 Functionality already exists

4 VOTE

allow to limit cpu usage to the number of slots requested

Hello,, currently, when a user submit a job, there's no way in LSF to limit the cpu consumption. recently, we discover that it's possible to limit cpu usage OS side using cgroup, by putting a value in the 'cpu.cfs_quota_us' file of the '/sys/fs/...

5 months ago in Spectrum LSF / Scheduling 0 Under review

2 VOTE

bjobs -o reserved_mem or bhosts -o reserved_mem

To see how much memory is reserved on a host, we need to parse the bhosts -l <hostname> command, and it doesn't give us info about which job is reserving how much memory. I'd either like to see one of the following: bhosts -o 'jobid reserved...

5 months ago in Spectrum LSF / Scheduling 1 Under review

3 VOTE

LSF connector error while configuring K8s

kubectl create -f $LSF_BINDIR/../../misc/kubernetes/parallelJob-v1alpha1.yaml > /tmp/IBM/parallelJob-v1alpha1.yaml.log error: resource mapping not found for name: "paralleljobs.ibm.com" namespace: "" from "/home/shared/lsfhome/10.1/linux3.10-...

6 months ago in Spectrum LSF / Scheduling 1 Future consideration

1 VOTE

Dynamic priority based on jobarray length

We exclusively work with jobarrays at our company. We have a queue reserved for jobs that take no more than 3 minutes. A general rule of thumb is that small jobarrays at our company can be considered as interactive jobs where the user waits for th...

6 months ago in Spectrum LSF / Scheduling 0 Not under consideration

1 VOTE

LSF does not allow jobs to moved from low to high/medium priority queues if there are limits set on medium/high priority queues

We have a setup as follows 1) High queue limit 3 2) Medium Queue limit 4 3) Low queue no limit and the default queue for usrs to submit jobs 4) Very low queue, will be used to use licenses effectively When a user submits 6 jobs for example in high...

7 months ago in Spectrum LSF / Scheduling 1 Future consideration

1 VOTE

Multi-cluster pre-exec failure should kill job from source cluster

These documents make it look like this issue is addressed, but they are not functioning as they should. https://www.ibm.com/docs/en/spectrum-lsf/10.1.0?topic=lsbparams-remote-max-preexec-retry https://www.ibm.com/docs/en/spectrum-lsf/10.1.0?topic=...

7 months ago in Spectrum LSF / Scheduling 0 Future 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...

9 months ago in Spectrum LSF / Scheduling 1 Not under consideration

Please enter your email address

FILTER BY CATEGORY