Admin Quick Start

This quick start gives an overview of installation of SingularityCE from source, a description of the architecture of SingularityCE, and pointers to configuration files. More information, including alternate installation options and detailed configuration options can be found later in this guide.

Architecture of SingularityCE

SingularityCE is designed to allow containers to be executed as if they were native programs or scripts on a host system. No daemon is required to build or run containers, and the security model is compatible with shared systems.

As a result, integration with clusters and schedulers such as Univa Grid Engine, Torque, SLURM, SGE, and many others is as simple as running any other command. All standard input, output, errors, pipes, IPC, and other communication pathways used by locally running programs are synchronized with the applications running locally within the container.

Beginning with SingularityCE 4 there are two modes in which to run containers:

  • The default native mode uses a runtime that is unique to SingularityCE. This is fully compatible with containers built for, and by, SingularityCE versions 2 and 3.

  • The optional OCI-mode uses a standard low-level OCI runtime to execute OCI containers natively, for improved compatibility.

SingularityCE favors an ‘integration over isolation’ approach to containers in native mode. By default only the mount namespace is isolated for containers, so that they have their own filesystem view. Default access to user home directories, /tmp space, and installation specific mounts makes it simple for users to benefit from the reproducibility of containerized applications without major changes to their existing workflows.

In OCI-mode, more isolation is used by default so that behavior is similar to Docker and other OCI runtimes. However, networking is not virtualized and SingularityCE’s traditional behavior can be enabled with the --no-compat option.

In both modes, access to hardware such as GPUs, high speed networks, and shared filesystems is easy and does not require special configuration. Where more complete isolation is important, SingularityCE can use additional Linux namespaces and other security and resource limits for that purpose.

SingularityCE Security

Note

See also the security section of this guide, for more detail.

When using the native runtime, a default installation of SingularityCE runs small amounts of privileged container setup code via a starter-setuid binary. This is a ‘setuid root’ binary, used so that SingularityCE can perform mounts, create namespaces, and enter containers even on older systems that lack support for fully unprivileged container execution. The setuid flow is the default mode of operation, but can be disabled upon build, or in the singularity.conf configuration file if required.

If setuid is disabled, or OCI-mode is used, SingularityCE sets up containers within an unprivileged user namespace. This makes use of features of newer kernels, as well as user space filesystem mounts (FUSE).

Note

Running SingularityCE in non-setuid mode requires unprivileged user namespace support in the operating system kernel and does not support all features. This impacts integrity/security guarantees of containers at runtime.

See the non-setuid installation section for further detail on how to install SingularityCE to run in non-setuid mode.

SingularityCE uses a number of strategies to provide safety and ease-of-use on both single-user and shared systems. Notable security features include:

  • When using the default native runtime, with container setup via a setuid helper program, the effective user inside a container is the same as the user who ran the container. This means access to files and devices from the container is easily controlled with standard POSIX permissions.

  • When using OCI-mode, or an unprivileged installation, subuid/subgid mappings allows users access to other uids & gids in the container, which map to safe administrator-defined ranges on the host.

  • Container filesystems are mounted nosuid and container applications run with the prctl NO_NEW_PRIVS flag set. This means that applications in a container cannot gain additional privileges. A regular user cannot sudo or otherwise gain root privilege on the host via a container.

  • The Singularity Image Format (SIF) supports encryption of containers, as well as cryptographic signing and verification of their content.

  • SIF containers are immutable and their payload is run directly, without extraction to disk. This means that the container can always be verified, even at runtime, and encrypted content is not exposed on disk.

  • Restrictions can be configured to limit the ownership, location, and cryptographic signatures of containers that are permitted to be run.

OCI Compatibility

SingularityCE allows users to run, and build from, the majority of OCI containers created with tools such as Docker. Beginning with SingularityCE 3.11, there are two modes of operation that support OCI containers in different ways.

SingularityCE’s native runtime, used by default, supports all features that are exposed via the singularity command. It builds and runs containers in SingularityCE’s own on-disk formats. When an OCI container is pulled or built into a SingularityCE image, a translation step occurs. While most OCI images are supported as-is, there are some limitations and compatibility options may be required.

SingularityCE 4’s OCI-mode, enabled with the --oci flag, runs containers using a low-level OCI runtime - either crun or runc. The container is executed from a native OCI format on-disk. Not all CLI features are currently implemented, but OCI containers using the USER directive or which are otherwise incompatible with SingularityCE’s native runtime are better supported. Note that OCI-mode has additional system requirements.

Version Compatibility

Up to and including version 4, the major version number of SingularityCE was increased only when a significant change or addition was made to the container format:

  • v1 packaged applications in a different manner than later versions, and was not widely deployed.

  • v2 used extfs or squashfs bare image files for the container root filesystem.

  • v3 introduced and switched to the Singularity Image Format (SIF).

  • v4 added OCI-SIF images, a variant of SIF encapsulating OCI containers directly. These are used by the new OCI-mode.

Minor versions, e.g. within the 3.x series, frequently introduced changes to existing behavior not related to the basic container format.

Beginning with version 4, SingularityCE aims to follow semantic versioning where breaking changes to the CLI or runtime behavior will also be limited to a new major version. New features that do not modify existing behavior may be introduced in minor version updates.

Backward Compatibility

Execution of container images from 2 prior major versions is supported. SingularityCE 4 can run container images created with versions 2 and 3. Except where documented in the project changelog, differences in behaviour when running v2 or v3 containers using the native runtime in setuid mode are considered bugs.

SingularityCE 4’s OCI-mode cannot perfectly emulate the behavior of the native runtime in setuid mode. Although most workflows are supported, complex containers created with SingularityCE 2 or 3 may not run as expected in OCI-mode.

Forward Compatibility

SingularityCE 4 can build SIF container images that can be run with version 3. The scope of this forward compatibility depends on the features used when creating the container, and the 3.x minor version used to run the container:

  • The OCI-SIF format (OCI-mode) is not supported before v4.

  • The SIF DSSE signature format (key / certificate based signing) was introduced at v3.11.0.

  • The SIF PGP signature format was changed at v3.6.0, therefore older versions cannot verify newer signatures.

  • Container / host environment handling was modified at v3.6.0.

  • LUKS2 encrypted containers are not supported prior to v3.4.0

SingularityCE 4.1 will build container images that can be run with version 4.0, with the exception of multi-layer OCI-SIF images. A multi-layer OCI-SIF created with the 4.1 --keep-layers option must be executed using version 4.1 or later.

Installation from Source

SingularityCE can be installed from source directly, or by building an RPM package from the source. Linux distributions may also package SingularityCE, but their packages may not be up-to-date with the upstream version on GitHub.

To install SingularityCE directly from source, follow the procedure below. Other methods are discussed in the Installation section.

Note

This quick-start that you will install as root using sudo, so that SingularityCE uses the default setuid workflow, and all features are available. See the non-setuid installation section of this guide for detail of how to install as a non-root user, and how this affects the functionality of SingularityCE.

Install Dependencies

On Debian-based systems, including Ubuntu:

# Ensure repositories are up-to-date
sudo apt-get update
# Install debian packages for dependencies
sudo apt-get install -y \
   autoconf \
   automake \
   cryptsetup \
   fuse \
   fuse2fs \
   git \
   libfuse-dev \
   libglib2.0-dev \
   libseccomp-dev \
   libtool \
   pkg-config \
   runc \
   squashfs-tools \
   squashfs-tools-ng \
   uidmap \
   wget \
   zlib1g-dev

On versions 8 or later of RHEL / Alma Linux / Rocky Linux, as well as on Fedora:

# Install basic tools for compiling
sudo yum groupinstall -y 'Development Tools'
# Install RPM packages for dependencies
sudo yum install -y \
   autoconf \
   automake \
   crun \
   cryptsetup \
   fuse \
   fuse3 \
   fuse3-devel \
   git \
   glib2-devel \
   libseccomp-devel \
   libtool \
   squashfs-tools \
   wget \
   zlib-devel

On version 7 of RHEL / CentOS:

# Install basic tools for compiling
sudo yum groupinstall -y 'Development Tools'
# Install RPM packages for dependencies
sudo yum install -y \
   autoconf \
   automake \
   cryptsetup \
   fuse \
   fuse3 \
   fuse3-devel \
   git \
   glib2-devel \
   libseccomp-devel \
   libtool \
   runc \
   squashfs-tools \
   wget \
   zlib-devel

On SLES / openSUSE Leap 15:

# Install RPM packages for dependencies
sudo zypper in \
   autoconf \
   automake \
   cryptsetup \
   fuse2fs \
   fuse3 \
   fuse3-devel \
   gcc \
   gcc-c++ \
   git \
   glib2-devel \
   libseccomp-devel \
   libtool \
   make \
   pkg-config \
   runc \
   squashfs \
   wget \
   zlib-devel

Note

You can build SingularityCE without cryptsetup available, but you will not be able to use encrypted containers without it installed on your system.

If you will not use the singularity oci commands, or OCI-mode, crun / runc is not required.

Install sqfstar / tar2sqfs for OCI-mode

If you intend to use the --oci execution mode of SingularityCE, your system must provide either:

  • squashfs-tools / squashfs >= 4.5, which provides the sqfstar utility. Older versions packaged by many distributions do not include sqfstar.

  • squashfs-tools-ng, which provides the tar2sqfs utility. This is not packaged by all distributions.

Debian / Ubuntu

On Debian/Ubuntu squashfs-tools-ng is available in the distribution repositories. It has been included in the “Install system dependencies” step above. No further action is necessary.

RHEL / Alma Linux / Rocky Linux / CentOS

On RHEL and derivatives, the squashfs-tools-ng package is now available in the EPEL repositories.

Follow the EPEL Quickstart for you distribution to enable the EPEL repository. Install squashfs-tools-ng with dnf or yum.

# EL 8 / 9
sudo dnf install squashfs-tools-ng

# EL 7
sudo yum install squashfs-tools-ng

SLES / openSUSE Leap

On SLES/openSUSE, follow the instructions at the filesystems project to obtain an more recent squashfs package that provides sqfstar.

Install Go

SingularityCE is written in Go, and you will need Go installed to compile it from source. Versions of Go packaged by your distribution may not be new enough to build SingularityCE.

{SingularityCE} aims to maintain support for the two most recent stable versions of Go. This corresponds to the Go Release Maintenance Policy and Security Policy, ensuring critical bug fixes and security patches are available for all supported language versions.

The method below is one of several ways to install and configure Go.

Note

If you have previously installed Go from a download, rather than an operating system package, you should remove your go directory, e.g. rm -r /usr/local/go before installing a newer version. Extracting a new version of Go over an existing installation can lead to errors when building Go programs, as it may leave old files, which have been removed or replaced in newer versions.

Visit the Go download page and pick a package archive to download. Copy the link address and download with wget. Then extract the archive to /usr/local (or use other instructions on go installation page).

$ export VERSION=1.20.4 OS=linux ARCH=amd64 && \
    wget https://dl.google.com/go/go$VERSION.$OS-$ARCH.tar.gz && \
    sudo tar -C /usr/local -xzvf go$VERSION.$OS-$ARCH.tar.gz && \
    rm go$VERSION.$OS-$ARCH.tar.gz

Finally, add /usr/local/go/bin to the PATH environment variable:

echo 'export PATH=$PATH:/usr/local/go/bin' >> ~/.bashrc
source ~/.bashrc

Download SingularityCE from a GitHub release

You can download SingularityCE from one of the releases. To see a full list, visit the GitHub release page. After deciding on a release to install, you can run the following commands to proceed with the installation.

$ export VERSION=main && # adjust this as necessary \
    wget https://github.com/sylabs/singularity/releases/download/v${VERSION}/singularity-ce-${VERSION}.tar.gz && \
    tar -xzf singularity-ce-${VERSION}.tar.gz && \
    cd singularity-ce-${VERSION}

Compile & Install SingularityCE

SingularityCE uses a custom build system called makeit. mconfig is called to generate a Makefile and then make is used to compile and install.

$ ./mconfig && \
    make -C ./builddir && \
    sudo make -C ./builddir install

By default SingularityCE will be installed in the /usr/local directory hierarchy. You can specify a custom directory with the --prefix option, to mconfig:

$ ./mconfig --prefix=/opt/singularity

This option can be useful if you want to install multiple versions of Singularity, install a personal version of SingularityCE on a shared system, or if you want to remove SingularityCE easily after installing it.

For a full list of mconfig options, run mconfig --help. Here are some of the most common options that you may need to use when building SingularityCE from source.

  • --sysconfdir: Install read-only config files in sysconfdir. This option is important if you need the singularity.conf file or other configuration files in a custom location.

  • --localstatedir: Set the state directory where containers are mounted. This is a particularly important option for administrators installing SingularityCE on a shared file system. The --localstatedir should be set to a directory that is present on each individual node.

  • -b: Build SingularityCE in a given directory. By default this is ./builddir.

  • --without-conmon: Do not build conmon, a container monitor that is used by the singularity oci commands. conmon is bundled with SingularityCE and will be built and installed by default. Use --without-conmon if you wish to use a version of conmon >=2.0.24 that is provided by your distribution rather than the bundled version. You can also specify --without-conmon if you know you will not use the singularity oci commands.

Installation from RPM/Deb Packages

Sylabs provides .rpm packages of SingularityCE, for mainstream-supported versions of RHEL and derivatives (e.g. Alma Linux / Rocky Linux). We also provide .deb packages for current Ubuntu LTS releases.

These packages can be downloaded from the GitHub release page and installed using your distribution’s package manager.

The packages are provided as a convenience for users of the open source project, and are built in our public CircleCI workflow. They are not signed, but SHA256 sums are provided on the release page.

Configuration

SingularityCE is configured using files under etc/singularity in your --prefix, or --syconfdir if you used that option with mconfig. In a default installation from source without a --prefix set you will find them under /usr/local/etc/singularity. In a default installation from RPM or Deb packages you will find them under /etc/singularity.

You can edit these files directly, or using the SingularityCE config global command as the root user to manage them.

singularity.conf contains the majority of options controlling the runtime behavior of SingularityCE. Additional files control security, network, and resource configuration. Head over to the Configuration files section where the files and configuration options are discussed.

Test SingularityCE

You can run a quick test of SingularityCE using a container in the Sylabs Container Library:

$ singularity exec library://alpine cat /etc/alpine-release
3.9.2

See the user guide for more information about how to use SingularityCE.