Commands: LSF → SLURM¶

This page maps common LSF commands to their SLURM equivalents.

Command mapping table¶

Action	LSF	SLURM	Notes
Submit a batch job	`bsub < script.sh`	`sbatch script.sh`	SLURM does not need input redirection
Interactive session	`bsub -Is -q interactive bash`	`srun --pty bash`	Add resource flags as needed
View job queue	`bjobs`	`squeue -u $USER`
View all jobs	`bjobs -a`	`squeue`
Detailed job info	`bjobs -l <jobid>`	`scontrol show job <jobid>`
Cancel a job	`bkill <jobid>`	`scancel <jobid>`
Cancel all my jobs	`bkill 0`	`scancel -u $USER`
Hold a job	`bstop <jobid>`	`scontrol hold <jobid>`
Release a held job	`bresume <jobid>`	`scontrol release <jobid>`
View cluster partitions	`bqueues`	`sinfo`
View partition details	`bqueues -l <queue>`	`sinfo -p <partition>`
Job history / accounting	`bhist <jobid>`	`sacct -j <jobid>`
Past job efficiency	—	`seff <jobid>`	SLURM-only; shows CPU/memory efficiency
View cluster load	`lsload`	`sinfo -N -l`
Node status	`bhosts`	`sinfo -N`
View job dependencies	`bjobs -l <jobid>`	`scontrol show job <jobid>`	Check `Dependency` field
Peek at job output	`bpeek <jobid>`	Read the `--output` file directly	No built-in equivalent
Modify pending job	`bmod`	`scontrol update job <jobid>`

Submitting jobs¶

LSFSLURM

# Submit a script
bsub < myjob.sh

# Submit with inline options
bsub -q normal -n 4 -W 2:00 -o out.%J < myjob.sh

# Submit with command directly
bsub -q short -o out.%J "echo hello"

# Submit a script
sbatch myjob.sh

# Submit with inline options
sbatch --partition=normal --ntasks=4 --time=02:00:00 --output=out.%j myjob.sh

# Submit with command directly (use --wrap)
sbatch --partition=short --output=out.%j --wrap="echo hello"

No input redirection needed

In LSF you pipe the script into bsub with <. In SLURM you pass the script filename as an argument to sbatch. If you forget and use sbatch < script.sh, it will still work — but the standard form is sbatch script.sh.

Monitoring jobs¶

LSFSLURM

# My running/pending jobs
bjobs

# Detailed info on a specific job
bjobs -l 12345

# All jobs in a queue
bjobs -q normal

# My running/pending jobs
squeue -u $USER

# Detailed info on a specific job
scontrol show job 12345

# All jobs in a partition
squeue -p normal

Job accounting¶

After a job completes, use sacct to review its resource usage:

# Basic accounting for a completed job
sacct -j 12345 --format=JobID,JobName,Partition,Elapsed,MaxRSS,State

# Quick efficiency summary
seff 12345

seff is your friend

seff <jobid> gives a quick summary of CPU and memory efficiency for completed jobs. Use it to right-size your future resource requests.