slurm is albedo's job scheduling system. It is used to submit jobs from the login nodes to the compute nodes.

Jobs

Account

To help attributing the usage of computing resources to the groups and projects of AWI, which is needed for reporting, on Albedo it is necessary to specify an account.
This is done by setting

-A, --account=<account>

...

sacctmgr modify user <user> set DefaultAccount=<account>

Partitions

Albedo’s compute nodes are divided into the following partitions, which are shown in the table below.

The smp partiton is the default and is for jobs which require less or equal than 128 cores. By default each core is attributed  256GB/128core = 2GB/core. Jobs in this partition might share a node.

Nodes in the mpp partition are exclusively reserved, and hence the entire memory is available for the job. This partition is used when one ore more nodes are needed.

The fat nodes can be selected via the fat partition. This partition resembles the smp partition but each node has much more memory.

Similarly, the GPU nodes can be accessed via the gpu partition. Note, that the type and number of GPUs need to be specified.

More infos about the hardware specification of each node can be found in the System Overview (TODO: Link).

Quality of service (QOS)

By default, the QOS 30min is used. It has a max. walltime of 30 minutes and jobs with this QOS get a higher priority and have access to a special SLURM reservation during working time (TODO: add details when set up), to facilitate development and testing. For longer runs, another QOS (and walltime) has to be specified. See table below. Note: long running jobs (longer than 12 hours, up to 48 hours) “cost” more in terms of fairshare.

A short note on the definitions:

UsageFactor: A float that is factored into a job’s TRES usage (e.g. RawUsage, …)

...

Submitting jobs

• To work interactively on a compute node use salloc.
  You can use all options (more CPU, RAM, time, partition, qos, ...) described in the next section.
  To enable working with graphical interfaces (X forwarding) add the option --x11 .
  
• Job scripts are submitted via sbatchYou can ssh to a node where a job of yours is running, if (and only if) you have a valid ssh-key pair. (e.g. on a login node: ssh-keygen -t ed25519;  ssh-copy-id albedo1)
  Make sure your key is secured wit ha password!

Specifying job resources

Job resources are defined at the header of your job script (or as command line arguments for sbatch or salloc). A full list see https://slurm.schedmd.com/sbatch.html#SECTION_OPTIONS. Here is a list of the most common ones:

...

 *) For filename patterns see: https://slurm.schedmd.com/sbatch.html#SECTION_%3CB%3Efilename-pattern%3C/B%3E

Scheduling

Priority

FILENAME-PATTERN

Job enforcements

We implemented some enforcements to improve albedo's overall performance.

• Jobs requesting --partition=fat but only low memory are rejected.
• Jobs requesting less than 40 nodes are enforced to use only nodes connected to the very same Infiniband switch (if this is feasible within 15 Minutes).

Details about specific parameters

Account (-A)

Compute resources are attributed to (primary)sections and projects at AWI. Therefore it is mandatory to specify an account.

The slurm accounts you may use are listed after login, with info.sh -s or can be shown via

sacctmgr -s show user name=$USER format=user,account%-30

Note: The account noaccount is just a dummy account that can not be used for computing.

You can change the default setting on your own:

sacctmgr modify user $USER set DefaultAccount=<account>

Partitions (-p)

Identical compute nodes are combined in partitions. More information about the hardware specification of each node can be found in the System Overview.

Quality of service (--qos)

Slurm's QOS is a way for us to influence a job's priority (priority QOS_factor) and "cost" (UsageFactor) based on the job's size (we only take walltime into account here!). We therefore created the different QOS, which are listed below.

The default QOS is 30min; for a job with a walltime >30min you have to select and set an appropriate QOS in addition to your walltime!

To facilitate development and testing, we have reserved 20 nodes during working hours exclusively for jobs with QOS=30min.

Job Scheduling

Priority

Jobs on albedo are scheduled based on a priority that is computed by Slurm depending on multiple factors (https://slurm.schedmd.com/priority_multifactor.html).
The higher the priority, the sooner your job beginsFor the job scheduling, Slurm assigns each job a priority, which is calculated based on several factors (Multifactor Priority Plugin). Jobs with higher priority, run first. (In principle – the backfill scheduling plugin helps making best use of available resources by filling up resources that are reserved (and thus idle) for large higher priority jobs with small (lower priority) jobs.)

On Albedo, Slurm is configured such, that the priority is mainly influenced by the Fairshare factor (which is based on the user’s recent use of resources; see Fairshare (TODO: Link)), while favoring short jobs (with qos=30min). With longer waiting time in the queue, a job’s priority increases. Job size, partitions or associations are not directly taken into account.

Fairshare

On Albedo all users have the same share of resources, independent of the account used. … TODO…

Accounting

TODO...

Information of jobs and nodes

...

At AWI, only few of the possible factors are taken into account:

Job_priority =   (PriorityWeightAge) * (age_factor)
                 + (PriorityWeightFairshare) * (fair-share_factor)
                 + (PriorityWeightQOS) * (QOS_factor)
                 - nice_factor

                 
The weights in this formula are set to balance the different factors and might become subject for tuning.
The current values can be assessed by running

$ scontrol show config | grep -i PriorityWeight
PriorityWeightAge       = 3500
PriorityWeightAssoc     = 0
PriorityWeightFairShare = 10000
PriorityWeightJobSize   = 0
PriorityWeightPartition = 0
PriorityWeightQOS       = 5000
PriorityWeightTRES      = (null)

The factors (except of the nice_factor (default is zero), which can be set by the user to downgrade the jobs priority by the setting --nice=...), are numbers in the range from 0 to 1.
They are shortly explained in the following.

You can check the recent usage of albedo with this command: 

sreport -t Percent cluster UserUtilizationByAccount  Start=$(date +%FT%T -d "1 week ago")  Format=used,login,account
FairShare

The fairshare factor is the most important factor here, but also the most difficult factor to understand. This factor is calculated using the "classic" fairshare algorithm of Slurm (https://slurm.schedmd.com/classic_fair_share.html). It computes the fairshare for each user based on the recent usage of the system.
Note, the usage of your associated account is *not* taken into accunt here, as it was the case on ollie!
Usage is basically "CPU seconds", but weighted using the UsageFactor depending on the used QOS (see section QOS). Furthermore, the usage taken into account here decays with time (with a half life time of 7 days).
Fairshare is the calculated by

FS = 2^(- (U_N / S_N) / D),

where the normalized usage U_N is the own usage relative to the total usage of albedo, the normalized share S_N is the share of a user on the entire system (1/(number of albedo users)) and D is a dampening factor. The formula basically assigns users a fairshare > 0,5 who under-use their share and < 0,5 for users who over-use their share. This is shown in the following figure, where the dots are taken from historic data from ollie. D has to be adjusted to account for the many users with an HPC account, who don't use it. This might also need tuning.

Image Added

Fairshare values can be shown with the command

sshare

QOS

To reward usage of the short 30min QOS for jobs, which are easier to schedule, the priority is increased!
See section about QOS.

Age

Job's priority slowly increases with waiting time in the queue. With the current setting the priority is increased by 500 for each day waiting. The factor saturates after 7 days.
Note: Jobs waiting for a dependency to finish are not ageing.

Useful Slurm commands

• sinfo shows existing queues
  For example to check how many nodes are available in a given partition (mpp, fat, gpu...)
  

  Code Block
  language bash
  sinfo -p<partition_name>

• scontrol show job <JobID> shows information about specific job
• sstat <JobID> shows resources used by a specific job
• squeue shows information about queues and used nodes
• sbatch <script> submits a batch job
• salloc <resources> requests access to compute nodes for interactive use
• scancel <JobID> cancels a batch job
• srun <ressources> <executable> starts a (parallel) code
• sshare and sprio give information on fair share value and job priority
• sreport -t Percent cluster UserUtilizationByAccount  Start=$(date +%FT%T -d "1 month ago")  Format=used,login,account | head -20 top usage users  during the last month

Do's & Don'ts

• Do not use srun for simple non-parallel jobs like cp, ln, rm, cat, g[un]zip
• Do not write loops in your slurm script to start several instance of similar jobs → See Job arrays below
• Make use of parallel srun p[gu]igz instead of g[un]zip if you have allocated more than one CPU already
• Do not allocate costly resources (like fat/gpu nodes) if you not need them. Check the CPU/Memory-Efficiency of your jobs with info.sh -S

Example Scripts

Job arrays

Job arrays in Slurm are an easy way to submit multiple similar jobs (e.g. executing the same script with multiple input data). See here for further details.

#!/bin/bash

#SBATCH --account=<account>          # Your account
#SBATCH --partition=smp
#SBATCH --time=0:10:00
#SBATCH --ntasks=1

# run 100 tasks, but only run 10 at a time
#SBATCH --array=1-100%10
#SBATCH --output=result_%A_%a.out    # gives result_<jobID>_<taskID>.out

echo "SLURM_JOBID:         $SLURM_JOBID"
echo "SLURM_ARRAY_TASK_ID: $SLURM_ARRAY_TASK_ID"
echo "SLURM_ARRAY_JOB_ID:  $SLURM_ARRAY_JOB_ID"

# Here we "translate" the $SLURM_ARRAY_TASK_ID (which takes values from 1-100)
# into an input file, that we want to analyze.
# Suppose 'input_files.txt' is a text file that has 100 lines, each containing
# the respective input file.

INPUT_LIST=input_files.txt

# Read the (SLURM_ARRAY_TASK_ID)th input file
INPUT_FILE=`sed -n "${SLURM_ARRAY_TASK_ID}p" < ${INPUT_LIST}`

srun my_executable $INPUT_FILE

MPI

#!/bin/bash

#SBATCH --account=<account>          # Your account 
#SBATCH --time=0:10:00
#SBATCH -p mpp
#SBATCH -N 2
#SBATCH --tasks-per-node=128
#SBATCH --cpus-per-task=1
#SBATCH --hint=nomultithread
#SBATCH --job-name=mpi
#SBATCH --output=out_%x.%j

# disable hyperthreading
#SBATCH --hint=nomultithread

module purge
module load    xthi/1.0-intel-oneapi-mpi2021.6.0-oneapi2022.1.0    intel-oneapi-mpi
# module load    xthi/1.0-openmpi4.1.3-gcc8.5.0   openmpi/4.1.3

## Uncomment the following line to enlarge the stacksize if needed,
##  e.g., if your code crashes with a spurious segmentation fault.
# ulimit -s unlimited

# To be on the safe side, we emphasize that it is pure MPI, no OpenMP threads
export OMP_NUM_THREADS=1

srun  xthi | sort -g -k 4

#!/bin/bash

#SBATCH --account=<account>          # Your account 
#SBATCH --time=0:10:00
#SBATCH -p mpp
#SBATCH -N 2
#SBATCH --tasks-per-node=31
#SBATCH --hint=nomultithread
#SBATCH --job-name=mpi_partial_node
#SBATCH --output=out_%x.%j

# disable hyperthreading
#SBATCH --hint=nomultithread

module purge
module load    xthi/1.0-intel-oneapi-mpi2021.6.0-oneapi2022.1.0   intel-oneapi-mpi
# module load    xthi/1.0-openmpi4.1.3-gcc8.5.0   openmpi/4.1.3

## Uncomment the following line to enlarge the stacksize if needed,
##  e.g., if your code crashes with a spurious segmentation fault.
# ulimit -s unlimited

# To be on the safe side, we emphasize that it is pure MPI, no OpenMP threads
export OMP_NUM_THREADS=1

# The --cpu-bind=rank_ldom distributes the tasks via the node's cores
# respecting the node's NUMA domains
srun --cpu-bind=rank_ldom xthi | sort -g -k 4

OpenMP

#!/bin/bash

#SBATCH --account=<account>          # Your account 
#SBATCH --time=0:10:00
#SBATCH -p smp
#SBATCH --tasks-per-node=1
#SBATCH --cpus-per-task=64
#SBATCH --job-name=openMP
#SBATCH --output=out_%x.%j

# disable hyperthreading
#SBATCH --hint=nomultithread

module purge
module load    xthi/1.0-intel-oneapi-mpi2021.6.0-oneapi2022.1.0   intel-oneapi-mpi
# module load    xthi/1.0-openmpi4.1.3-gcc8.5.0   openmpi/4.1.3

## Uncomment the following line to enlarge the stacksize if needed,
##  e.g., if your code crashes with a spurious segmentation fault.
# ulimit -s unlimited
# export OMP_STACKSIZE=128M

# This binds each thread to one core
export OMP_PROC_BIND=TRUE

# OpenMP and srun, both need to know the number of CPUs per task
export OMP_NUM_THREADS=$SLURM_CPUS_PER_TASK
export SRUN_CPUS_PER_TASK=$SLURM_CPUS_PER_TASK

srun xthi | sort -g -k 4

Hybrid (MPI+OpenMP)

#!/bin/bash

#SBATCH --account=<account>          # Your account 
#SBATCH --time=0:10:00
#SBATCH -p mpp
#SBATCH -N 2
#SBATCH --tasks-per-node=8
#SBATCH --cpus-per-task=16
#SBATCH --job-name=hybrid
#SBATCH --output=out_%x.%j

# disable hyperthreading
#SBATCH --hint=nomultithread

module purge
module load    xthi/1.0-intel-oneapi-mpi2021.6.0-oneapi2022.1.0   intel-oneapi-mpi
# module load    xthi/1.0-openmpi4.1.3-gcc8.5.0   openmpi/4.1.3

## Uncomment the following line to enlarge the stacksize if needed,
##  e.g., if your code crashes with a spurious segmentation fault.
# ulimit -s unlimited
# export OMP_STACKSIZE=128M

# This binds each thread to one core
export OMP_PROC_BIND=TRUE

# OpenMP and srun, both need to know the number of CPUs per task
export OMP_NUM_THREADS=$SLURM_CPUS_PER_TASK
export SRUN_CPUS_PER_TASK=$SLURM_CPUS_PER_TASK

srun xthi | sort -g -k 4

Usage of GPU

#!/bin/bash

#SBATCH --account=<account>          # Your account 
#SBATCH --time=0:10:00
#SBATCH -p gpu
#SBATCH --ntasks=1
#SBATCH --gpus=a100:2                 # allocate 2 (out of 4) A100 GPUs; to get 2 (out of 2) A40 GPUs use --gpus=a40:2
#SBATCH --hint=nomultithread
#SBATCH --job-name=gpu
#SBATCH --output=out_%x.%j

# disable hyperthreading
#SBATCH --hint=nomultithread

## Uncomment the following line to enlarge the stacksize if needed,
##  e.g., if your code crashes with a spurious segmentation fault.
# ulimit -s unlimited

# To be on the safe side, we emphasize that it is pure MPI, no OpenMP threads
export OMP_NUM_THREADS=1

srun your_code_that_runs_on_GPUs