# FAQ
For example problems
# How do I log on to machines on Linux?
Systems from the UNIX/Linux family usually have programs that allow SSH connections. To log in you need to execute a command:
ssh <login>@<company_machine_name>
where <login> is the name of the user account on the machine and <access_machine name> is the name of the access machine or its IP number.
For example, a user with a novak account on a eagle.man.poznan.pl machine will log in using a command:
ssh nowak@eagle.man.poznan.pl
# How do I log on to machines in Windows?
Windows systems do not have a built-in SSH protocol client. Therefore their users must install additional software.
Recommended programs:
Putty (opens new window) as a client of SSH
WinSCP (opens new window) for file transfer
After logging in to the machine, the user uses the text console, just like when logging in from Unix/Linux machines.
Sample script
#!/bin/bash
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=4
#SBATCH --mem=4gb
#SBATCH --time=01:00:00
# We set up paths or load appropriate modules
module load plink/1.90
# Set the $TMPDIR variable
export TMPDIR=$HOME/grant_$SLURM_JOB_ACCOUNT/scratch/$USER/$SLURM_JOB_ID
# We set application variables
export SCR=${TMPDIR}
# We set the auxiliary variables
INPUT_DIR="input"
OUTPUT_DIR="output"
OUTPUT_FILE="OUTPUT"
# We create a temporary directory
mkdir -p ${TMPDIR}
# Copy the input data to the directory indicated for the variable $TMPDIR
cp ${SLURM_SUBMIT_DIR}/${INPUT_DIR}/* ${TMPDIR}
# We go to the $TMPDIR directory
cd $TMPDIR
# We make calculations
plink --noweb --file hapmap1
# We finish calculations, copy the contents of the $TMPDIR/output directory
# to the directory from which the task was sealed.
mkdir $SLURM_SUBMIT_DIR/${OUTPUT_DIR}
cp -r $TMPDIR/* $SLURM_SUBMIT_DIR/${OUTPUT_DIR}/
# We're cleaning the working directory
rm -rf $TMPDIR
# How to become a user
For details, see How to become a user (opens new window).
# Can I start calculation tasks on the access machine?
It is not permitted to start computation tasks on an access node. Any such task will be interrupted without warning. Tasks can only be started with a queuing system.
# How to run a job on more than one processor?
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=4
# How do you know how many cores a cluster node has?
You can find out the number of cores in a cluster node using the scontrol show node command in SLURM (Simple Linux Utility for Resource Management). This command provides detailed information about each node, including its hardware and configuration specifics.
Here's a breakdown of how to interpret the core-related details in the output:
Command:
scontrol show node <node-name>
Replace
# How do I request XXXX megabytes of memory for my task?
In the queuing system script, you should put a line that specifies the memory size. For 2GB it has a form:
#SBATCH --mem=2gb
# How do you specify the exact time that the task will count?
The time parameter is used to specify the exact time needed for the task. The time is given in the format HH:MM:SS. Use in a queueing script (a task that will last 12 hours and 30 minutes):
#SBATCH --time=12:30:00
# Why doesn't my job start right away despite my R status in the queue system
The behaviour of the queuing system is correct and results from the fact that the queuing system has been configured in such a
way as to minimize the energy consumption of the cluster. This mechanism causes nodes that are not used by the queuing system
to be disabled. When the system accepts the task, the nodes are restarted. Switching on a calculation node takes a maximum of
5 minutes and during this time the queuing system shows that the task is running (R) and has the nodes assigned, that are just
starting up (not yet available through ssh). In such a situation the calculations should start after maximum 5 minutes.
In case one of the nodes is not started correctly, the queuing system should redirect the task eliminating the malfunctioning nodes.
# Do you offer resources via CLI (command line interface)?
Yes, users can access our system resources through the CLI (command line interface). Access is provided via SSH, which
allows connecting to servers and performing computations in a Linux environment. Users can use popular SSH clients,
such as standard terminal tools (PuTTY, WinSCP), or web-based SSH applications.
# What is the maximum amount of resources one can apply for?
We do not have fixed limits for individual users. When applying, we offer predefined grants for convenience,
but users can also specify their needs individually. Each application is reviewed by experts who assess whether the requested
resources are suitable for the intended project. Once approved, resources are allocated as requested. Details on how to apply
can be found here: How to Become a User of the New Portal (opens new window).
# What operating system and queuing system are used?
Currently, we are using CentOS Linux version 7, and SLURM serves as the queuing system for our cluster.
# Can users create their own Conda environments or use Docker/Singularity containers?
Yes, users can create their own Conda environments and use Singularity (Apptainer) containers to install the necessary software.
Docker is restricted due to root access requirements, but Docker images can be run within the Singularity environment.
SLURM has good support for containerized workflows.
# Are compatible versions of CUDA and cuDNN installed on GPU clusters?
Our clusters support two types of GPUs: older V100s and newer H100s. We currently support CUDA version 12.0.1 and periodically
update to the latest versions compatible with our drivers. Older and custom versions of CUDA can also be used in
Conda environments and containers.
# What storage space is available for individual users?
Each user has a default 1 GB limit in their home directory. After obtaining a grant, a dedicated project directory is created with
links to different types of storage, such as archive, project data, or scratch space. Limits vary depending on the data type;
large projects can receive over 1 PB of storage.