Next monthly maintenance November 1st [DETAILS]

Search Docs by Keyword

Singularity on the cluster

Table of Contents:


Containerization of workloads has become popular, particularly using Docker. However, Docker is not suitable for HPC applications due to security reasons. There are a couple of alternatives for HPC containers, with Singularity being the one that covers a large set of cases. Singularity has been deployed on the cluster, and can also import Docker containers.
This page provides information on how to use Singularity on the cluster. Singularity enables users to have full control of their operating system environment. This allows a non-privileged user to “swap out” the Linux operating system and environment on the host machine for a Linux OS and computing environment that they can control. For instance, if the host system runs CentOS Linux but your application requires Ubuntu Linux with a specific software stack, you can create an Ubuntu image, install your software into that image, copy the created image to the cluster, and run your application on that host in its native Ubuntu environment.
Singularity leverages the resources of the host system, such as high-speed interconnect (e.g., InfiniBand), high-performance parallel file systems (e.g., Lustre /n/holyscratch01 and /n/holylfs filesystems), GPUs, and other resources (e.g., licensed Intel compilers).
Note for Windows and MacOS: Singularity only supports Linux containers. You cannot create images that use Windows or MacOS (this is a restriction of the containerization model rather than Singularity).

Why Singularity?

There are some important differences between Docker and Singularity:

  • Docker and Singularity have their own container formats.
  • Docker containers may be imported to run via Singularity.
  • Docker containers need root privileges for full functionality which is not suitable for a shared HPC environment.
  • Singularity allows working with containers as a regular user.

Singularity on the cluster

Singularity is available only on the compute nodes on the cluster. Therefore, to use it you need to either start an interactive job or submit a batch-job to the available SLURM queues.
In the below examples we illustrate the interactive use of Singularity in an interactive bash shell.

[user@holylogin01 ~]$ salloc -p test -c 1 -t 00-01:00 --mem=4000
[user@holyseas02 ~]$

Check Singularity version:

[user@holyseas02 ~]$ which singularity
[user@holyseas02 ~]$ singularity --version
singularity version 3.1.1-1.el7

The most up-to-date help on Singularity comes from the command itself.

[user@holyseas02 ~]$ singularity --help
Linux container platform optimized for High Performance Computing (HPC) and
Enterprise Performance Computing (EPC)
  singularity [global options...]
  Singularity containers provide an application virtualization layer enabling
  mobility of compute via both application and environment portability. With
  Singularity one is capable of building a root file system that runs on any
  other Linux system where Singularity is installed.
  -d, --debug              print debugging information (highest verbosity)
  -h, --help               help for singularity
  -q, --quiet              suppress normal output
  -s, --silent             only print errors
  -t, --tokenfile string   path to the file holding your sylabs
                           authentication token (default
  -v, --verbose            print additional information
      --version            version for singularity
Available Commands:
  apps        List available apps within a container
  build       Build a Singularity image
  cache       Manage the local cache
  capability  Manage Linux capabilities for users and groups
  exec        Run a command within a container
  help        Help about any command
  inspect     Show metadata for an image
  instance    Manage containers running as services
  key         Manage OpenPGP keys
  oci         Manage OCI containers
  pull        Pull an image from a URI
  push        Push a container to a Library URI
  run         Run the user-defined default command within a container
  run-help    Show the user-defined help for an image
  search      Search a Library for images
  shell       Run a shell within a container
  sign        Attach a cryptographic signature to an image
  test        Run the user-defined tests within a container
  verify      Verify cryptographic signatures attached to an image
  version     Show the version for Singularity
  $ singularity help 
      Additional help for any Singularity subcommand can be seen by appending
      the subcommand name to the above command.

For additional help or support, please visit

Getting existing images onto the cluster

Singularity uses container images which you can scp or rsync to the cluster as you would do with any other file. See Copying Data to & from the cluster using SCP or SFTP for more information.
Note: For larger Singularity images, please use the available scratch filesystems, such as /n/holyscratch01/my_lab/username and /n/holylfs/LABS/my_lab/username.
You can also use the pull or build commands to download pre-built images from external resources, such as Singularity Hub, the Sylabs Container Library or Docker Hub. For instance, you can download a native Singularity image with its default name from Singularity Hub with:

[user@holyseas02 ~]$ singularity pull shub://vsoch/hello-world
 62.32 MiB / 62.32 MiB [================================] 100.00% 43.02 MiB/s 1

The downloaded image file is hello-world_latest.sif.
You can also pull the image with a customized name; e.g., hello.sif:

[user@holyseas02 ~]$ singularity pull --name hello.sif shub://vsoch/hello-world
 62.32 MiB / 62.32 MiB [================================] 100.00% 57.26 MiB/s 1s

Similarly, you can pull images from Docker Hub:

[user@holyseas02 ~]$ singularity pull docker://godlovedc/lolcow
INFO:    Starting build...
Getting image source signatures
Copying blob sha256:9fb6c798fa41e509b58bccc5c29654c3ff4648b608f5daa67c1aab6a7d02c118
Writing manifest to image destination
Storing signatures
INFO:    Creating SIF file...
INFO:    Build complete: lolcow_latest.sif

See official Singularity documentation for more information.


Cluster nodes automount a CernVM-File System at /cvmfs/ This provides a universal file system namespace to Singularity images for the BioContainers project, which comprises container images automatically generated from Bioconda software packages. The Singularity images are organized into a directory hierarchy following the convention:


For example:

singularity exec /cvmfs/ samtools --help

The Bioconda package index lists all software available in /cvmfs/, while the BioContainers registry provides a searchable interface.

NOTE: There will be a 10-30 second delay when first accessing /cvmfs/ on a compute node on which it is not currently mounted; in addition, there will be a delay when accessing a Singularity image on a compute node where it has not already been accessed and cached to node-local storage.

Docker Rate Limiting

Docker rate limits the number of pulls anonymous accounts can make from Docker Hub.  If you hit either an error of Too Many Requests or pull rate limit, you will need to create a Docker account to get a higher limit.  See the Docker documentation for more details.

Working with images

When working with images you could either start an interactive session, or submit a Singularity job to the available queues. For these examples, we will use a hello-world.sif in an interactive bash shell.

[user@holylogin01 ~]$ salloc -p test -c 1 -t 00-01:00 --mem=4000
[user@holyseas02 ~]$ singularity pull --name hello-world.sif shub://vsoch/hello-world
 62.32 MiB / 62.32 MiB [================================] 100.00% 37.63 MiB/s 1s


With the shell command, you can start a new shell within the container image and interact with it as if it were a small virtual machine.

[user@holyseas02 ~]$ singularity shell hello-world.sif
Singularity hello-world.sif:~/holylfs/pgk/SINGULARITY/vol2> pwd
Singularity hello-world.sif:~/holylfs/pgk/SINGULARITY/vol2> ls
funny.sif  gcc-7.2.0.sif  hello-world.sif  hello.sif  lolcow.sif  ubuntu.sif  vsoch-hello-world-master-latest.sif
Singularity hello-world.sif:~/holylfs/pgk/SINGULARITY/vol2> id
uid=56139(pkrastev) gid=40273(rc_admin) groups=40273(rc_admin),10006(econh11),34539(fas_it),34540(cluster_users),402119(solexa_writers),402160(VPN_HELPMAN),402161(RT_Users),402854(wpdocs_users),403083(owncloud),403266(file-isi_microsoft-full-dlg),403284(gitlabint_users),403331(rc_class)
Singularity hello-world.sif:~/holylfs/pgk/SINGULARITY/vol2>

Commands within a container

You can use the exec command to execute specific commands within the container. For instance, you can run the below command to display information about the native Linux OS of the image:

[user@holyseas02 ~]$ singularity exec hello-world.sif cat /etc/os-release
VERSION="14.04.5 LTS, Trusty Tahr"
PRETTY_NAME="Ubuntu 14.04.5 LTS"

Running containers

Singularity images contain run-scripts that can be triggered with the run command to perform specific actions when the container is run. This can be done either by using the run command, or by calling the container as if it were an executable, i.e.,

[user@holyseas02 ~]$ singularity run hello-world.sif


[user@holyseas02 ~]$ ./hello-world.sif

Sometimes you may have a container with several apps, each with its own set of run-scripts. You can use the apps command to list the available apps within the container. For instance, if you have an image named my_image.simg which has N apps (app_1, app_2,…, app_N) you can do:

[user@holyseas02 ~]$ singularity apps my_image.sif

You can run a particular app with

[user@holyseas02 ~]$ singularity run --app app_2 my_image.sif

GPU example:

To access Nvidia GPU card driver installed inside of Singularity container you need to use –nv option while executing the container.

[user@holylogin01 ~]$ salloc -p gpu --gres=gpu:1 --mem 1000 -n 4 -t 600
[user@supermicgpu01 ~]$ singularity pull --name hello-world.sif shub://vsoch/hello-world
[user@supermicgpu01 ~]$ singularity exec --nv hello-world.sif /bin/bash
[user@supermicgpu01 ~]$ nvidia-smi
| NVIDIA-SMI 396.26 Driver Version: 396.26 |
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| 0 Tesla K20Xm Off | 00000000:88:00.0 Off | 0 |
| N/A 37C P0 61W / 235W | 0MiB / 5700MiB | 65% Default |

To verify, if you have access to the requested GPUs, run “nvidia-smi”

Accessing files from a container

Files and directories on the cluster are accessible from within the container. By default, directories under /n, $HOME, $PWD, and /tmp are available at runtime.
You can specify additional directories to bind mount into your container with the --bind option. For instance, in the below example the /scratch directory on the host system is bind mounted to the /mnt directory inside the container:

[user@holyseas02 ~]$ echo 'Hello from inside the container!' > /scratch/hello.dat
[user@holyseas02 ~]$ singularity exec --bind /scratch:/mnt hello-world.sif cat /mnt/hello.dat
Hello from inside the container!

Singularity containers as SLURM jobs

You can also use Singularity images within a non-interactive batch script as you would any other command. If your image contains a run-script then you can use singularity run to execute the run-script in the job. You can also use singularity exec to execute arbitrary commands (or scripts) within the image. Below is an example batch-job submission script using the hello-world.simg to print out information about the native OS of the image.

#SBATCH -J singularity_test
#SBATCH -o singularity_test.out
#SBATCH -e singularity_test.err
#SBATCH -p shared
#SBATCH -t 0-00:30
#SBATCH -c 1
#SBATCH --mem=4000
# Singularity command line options
singularity exec hello-world.sif cat /etc/os-release

If the above batch-job script is named singularity.sbatch, for instance, the jobs is submitted as usual with sbatch:

[user@holylogin01 ~]$ sbatch singularity.sbatch

Upon the job completion, the STD output is located in the file singularity_test.out.

[user@holylogin01 ~]$ cat singularity_test.out
VERSION="14.04.5 LTS, Trusty Tahr"
PRETTY_NAME="Ubuntu 14.04.5 LTS"

Building Singularity images

To build Singularity containers, you need root access to the build system. Therefore, you cannot build a Singularity container on the cluster. Depending on whether or not you have an access to a Linux machine, possible options are:

  • If you have a Linux system to which you have root (admin) access, you can install Singularity and build your Singularity containers there. See Install Singularity on Linux for more information.
  • If you don’t have a Linux system you could easily install one in a virtual machine using software like VirtualBox, Vagrant, or VMware. See Install on Windows or Mac for specific Windows or MacOS instructions.
  • You can build a Singularity image from a local Docker image on Windows, Mac, or Linux with the docker2singularity Docker image.

In addition to your own Linux environment, you will also need a definition file to build a Singularity container from scratch. You can find some simple definition files for a variety of Linux distributions in the /example directory of the source code. Detailed documentation about building Singularity container images is available at the Singularity website.


© The President and Fellows of Harvard College
Except where otherwise noted, this content is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International license.