Search Docs by Keyword
Running Singularity image with CentOS 7 on Rocky 8
If you absolutely need to still use CentOS 7 after the OS upgrade to Rocky 8, you can use it with SingularityCE. For more information on SingularityCE, see our Singularity documentation and GitHub User Codes.
We have a Singularity image with CentOS7 and the same environment of compute nodes (as of March 29th, 2023). In addition, you can access CentOS 7 modules from within the Singularity container. The image is stored at:
/n/singularity_images/FAS/centos7/compute-el7-noslurm-2023-03-29.sif
You can execute this image and/or copy it, but you cannot modify it in its original location. See below how you can modify this image.
Run Singularity with CentOS 7
To get a bash shell on CentOS 7 environment, you can run:
[jharvard@holy7c12102 ~]$ singularity run /n/singularity_images/FAS/centos7/compute-el7-noslurm-2023-03-29.sif
Singularity>
or
[jharvard@holy7c12102 ~]$ singularity exec /n/singularity_images/FAS/centos7/compute-el7-noslurm-2023-03-29.sif /bin/bash
Singularity>
NOTE: The command singularity shell is also an option. However it does not allow you to access modules as explained in Load modules
Load modules
You can still load modules that were available on CentOS 7 from inside the Singularity container:
[jharvard@holy7c12102 ~]$ singularity exec /n/singularity_images/FAS/centos7/compute-el7-noslurm-2023-03-29.sif /bin/bash
Singularity> module load gcc
Singularity> module load matlab
Singularity> module list
Currently Loaded Modules:
1) gmp/6.2.1-fasrc01 2) mpfr/4.1.0-fasrc01 3) mpc/1.2.1-fasrc01 4) gcc/12.1.0-fasrc01 5) matlab/R2022b-fasrc01
Note that the modules are from the CentOS 7 environment:
Singularity> module display matlab/R2022b-fasrc01
-----------------------------------------------------------------------------------------------------------------------------------------------------------
/n/helmod/modulefiles/centos7/Core/matlab/R2022b-fasrc01.lua:
-----------------------------------------------------------------------------------------------------------------------------------------------------------
help([[matlab-R2022b-fasrc01
a high-level language and interactive environment for numerical computation, visualization, and programming
]], [[
]])
whatis("Name: matlab")
whatis("Version: R2022b-fasrc01")
whatis("Description: a high-level language and interactive environment for numerical computation, visualization, and programming")
setenv("MATLAB_HOME","/n/helmod/apps/centos7/Core/matlab/R2022b-fasrc01")
prepend_path("PATH","/n/helmod/apps/centos7/Core/matlab/R2022b-fasrc01/bin")
setenv("MLM_LICENSE_FILE","27000@rclic1")
setenv("ZIENA_LICENSE_NETWORK_ADDR","10.242.113.134:8349")
Submit slurm jobs
If you need to submit a job rather than getting to a shell, you have to do the following steps in the appropriate order:
- launch the singularity image
- load modules
- (optional) compile code
- execute code
If you try to load modules before launching the image, it will try to load modules from the Rocky 8 host system.
To ensure that steps 2-4 are run within the singularity container, they are place between END (see slurm batch script below).
NOTE: You cannot submit slurm jobs from inside the container, but you can submit a slurm job that will execute the container.
Example with a simple hello_world.f90 fortran code:
program hello
print *, 'Hello, World!'
end program hello
Slurm batch script run_singularity_centos7.sh:
#!/bin/bash
#SBATCH -J sing_hello # Job name
#SBATCH -p rocky # Partition(s) (separate with commas if using multiple)
#SBATCH -c 1 # Number of cores
#SBATCH -t 0-00:10:00 # Time (D-HH:MM:SS)
#SBATCH --mem=500M # Memory
#SBATCH -o sing_hello_%j.out # Name of standard output file
#SBATCH -e sing_hello_%j.err # Name of standard error file
# start a bash shell inside singularity image
singularity run /n/singularity_images/FAS/centos7/compute-el7-noslurm-2023-03-29.sif <<END
# load modules
module load gcc/12.1.0-fasrc01
module list
# compile code
gfortran hello_world.f90 -o hello.exe
# execute code
./hello.exe
END
To ensure that the commands are run within the singularity container, they are place between END.
To submit the slurm batch script:
sbatch run_singularity_centos7.sh
Another option have a bash script with steps 2-4 and then use singularity run to execute the script. For example, script_inside_container.sh:
#!/bin/bash
# load modules
module load gcc/12.1.0-fasrc01
module list
# compile code
gfortran hello_world.f90 -o hello.exe
# execute code
./hello.exe
And the slurm batch script run_singularity_centos7_script.sh becomes:
#!/bin/bash
#SBATCH -J sing_hello # Job name
#SBATCH -p rocky # Partition(s) (separate with commas if using multiple)
#SBATCH -c 1 # Number of cores
#SBATCH -t 0-00:10:00 # Time (D-HH:MM:SS)
#SBATCH --mem=500M # Memory
#SBATCH -o sing_hello_%j.out # Name of standard output file
#SBATCH -e sing_hello_%j.err # Name of standard error file
# start a bash shell inside singularity image
singularity run /n/singularity_images/FAS/centos7/compute-el7-noslurm-2023-03-29.sif script_inside_container.sh
You can submit a batch job with:
sbatch run_singularity_centos7_script.sh
Modify SingularityCE image with CentOS 7
If you need to run your codes in the former operating system (pre June 2023), you can build a custom SingularityCE image with proot. The base image is the the FASRC CentOS 7 compute node image, and you can add your own software/library/packages under the %post header in the Singularity definition file.
Step 1: Follow steps 1 and 2 in setup proot
Step 2: Copy the CentOS 7 compute image to your holylabs (or home directory). The base image file needs to be copied to a directory that you have read/write access, otherwise it will fail to build your custom image
[jharvard@holy2c02302 ~]$ cd /n/holylabs/LABS/jharvard_lab/Users/jharvard
[jharvard@holy2c02302 jharvard]$ cp /n/holystore01/SINGULARITY/FAS/centos7/compute-el7-noslurm-2023-02-15.sif .
Step 3: In definition file centos7_custom.def, set Bootstrap: localimage and put the path of the existing image that you copied for the From: field. Then, add your packages/software/libraries that you need. Here, we add cowsay:
Bootstrap: localimage
From: compute-el7-noslurm-2023-02-15.sif
%help
This is CentOS 7 Singularity container based on the Cannon compute node with my added programs.
%post
yum -y update
yum -y install cowsay
Step 3: Build the SingularityCE image
[jharvard@holy2c02302 jharvard]$ singularity build centos7_custom.sif centos7_custom.def
INFO: Using proot to build unprivileged. Not all builds are supported. If build fails, use --remote or --fakeroot.
INFO: Starting build...
INFO: Verifying bootstrap image compute-el7-noslurm-2023-02-15.sif
WARNING: integrity: signature not found for object group 1
WARNING: Bootstrap image could not be verified, but build will continue.
INFO: Running post scriptlet
+ yum -y update
... omitted output ...
Running transaction
Installing : cowsay-3.04-4.el7.noarch 1/1
Verifying : cowsay-3.04-4.el7.noarch 1/1
Installed:
cowsay.noarch 0:3.04-4.el7
Complete!
INFO: Adding help info
INFO: Creating SIF file...
INFO: Build complete: centos7_custom.sif