Containers Guide

Containers are being adopted in HPC workloads. Containers rely on existing kernel features to allow greater user control over what applications see and can interact with at any given time. For HPC Workloads, these are usually restricted to the mount namespace. Slurm natively supports the requesting of unprivileged OCI Containers for jobs and steps.

Known limitations

The following is a list of known limitations of the Slurm OCI container implementation.

  • All containers must run under unprivileged (i.e. rootless) invocation. All commands are called by Slurm as the user with no special permissions.
  • Custom container networks are not supported. All containers should work with the "host" network.
  • Slurm will not transfer the OCI container bundle to the execution nodes. The bundle must already exist on the requested path on the execution node.
  • Containers are limited by the OCI runtime used. If the runtime does not support a certain feature, then that feature will not work for any job using a container.
  • oci.conf must be configured on the execution node for the job, otherwise the requested container will be ignored by Slurm (but can be used by the job or any given plugin).

Prerequisites

The host kernel must be configured to allow user land containers:

$ sudo sysctl -w kernel.unprivileged_userns_clone=1

Docker also provides a tool to verify the kernel configuration: 

$ dockerd-rootless-setuptool.sh check --force
[INFO] Requirements are satisfied

Required software:

  • Fully functional OCI runtime. It needs to be able to run outside of Slurm first.
  • Fully functional OCI bundle generation tools. Slurm requires OCI Container compliant bundles for jobs.

Example configurations for various OCI Runtimes

The OCI Runtime Specification provides requirements for all compliant runtimes but does not expressly provide requirements on how runtimes will use arguments. In order to support as many runtimes as possible, Slurm provides pattern replacement for commands issued for each OCI runtime operation. This will allow a site to edit how the OCI runtimes are called as needed to ensure compatibility.

For runc and crun, there are two sets of examples provided. The OCI runtime specification only provides the start and create operations sequence, but these runtimes provides a much more efficient run operation. Sites are strongly encouraged to use the run operation (if provided) as the start and create operations require that Slurm poll the OCI runtime to know when the containers have completed execution. While Slurm attempts to be as efficient as possible with polling, it will result in a thread using CPU time inside of the job and slower response of Slurm to catch when container execution is complete.

The examples provided have been tested to work but are only suggestions. Sites are expected to ensure that the resultant root directory used will be secure from cross user viewing and modifications. The examples provided point to "/run/user/%U" where %U will be replaced with the numeric user id which should be created and managed by systemd independently of Slurm.

  • runc using create/start: 
    EnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeEnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeQuery="runc --rootless=true --root=/run/user/%U/ state %n.%u.%j.%s.%t"
RunTimeCreate="runc --rootless=true --root=/run/user/%U/ create %n.%u.%j.%s.%t -b %b"
RunTimeStart="runc --rootless=true --root=/run/user/%U/ start %n.%u.%j.%s.%t"
RunTimeKill="runc --rootless=true --root=/run/user/%U/ kill -a %n.%u.%j.%s.%t"
RunTimeDelete="runc --rootless=true --root=/run/user/%U/ delete --force %n.%u.%j.%s.%t"
  • runc using run (suggested): 
    EnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeEnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeQuery="runc --rootless=true --root=/run/user/%U/ state %n.%u.%j.%s.%t"
RunTimeKill="runc --rootless=true --root=/run/user/%U/ kill -a %n.%u.%j.%s.%t"
RunTimeDelete="runc --rootless=true --root=/run/user/%U/ delete --force %n.%u.%j.%s.%t"
RunTimeRun="runc --rootless=true --root=/run/user/%U/ run %n.%u.%j.%s.%t -b %b"
  • crun using create/start: 
    EnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeEnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeQuery="crun --rootless=true --root=/run/user/%U/ state %n.%u.%j.%s.%t"
RunTimeKill="crun --rootless=true --root=/run/user/%U/ kill -a %n.%u.%j.%s.%t"
RunTimeDelete="crun --rootless=true --root=/run/user/%U/ delete --force %n.%u.%j.%s.%t"
RunTimeCreate="crun --rootless=true --root=/run/user/%U/ create --bundle %b %n.%u.%j.%s.%t"
RunTimeStart="crun --rootless=true --root=/run/user/%U/ start %n.%u.%j.%s.%t"
  • crun using run (suggested): 
    EnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeEnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeQuery="crun --rootless=true --root=/run/user/%U/ state %n.%u.%j.%s.%t"
RunTimeKill="crun --rootless=true --root=/run/user/%U/ kill -a %n.%u.%j.%s.%t"
RunTimeDelete="crun --rootless=true --root=/run/user/%U/ delete --force %n.%u.%j.%s.%t"
RunTimeRun="crun --rootless=true --root=/run/user/%U/ run --bundle %b %n.%u.%j.%s.%t"
  • nvidia-container-runtime using create/start: 
    EnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeEnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeQuery="nvidia-container-runtime --rootless=true --root=/run/user/%U/ state %n.%u.%j.%s.%t"
RunTimeCreate="nvidia-container-runtime --rootless=true --root=/run/user/%U/ create %n.%u.%j.%s.%t -b %b"
RunTimeStart="nvidia-container-runtime --rootless=true --root=/run/user/%U/ start %n.%u.%j.%s.%t"
RunTimeKill="nvidia-container-runtime --rootless=true --root=/run/user/%U/ kill -a %n.%u.%j.%s.%t"
RunTimeDelete="nvidia-container-runtime --rootless=true --root=/run/user/%U/ delete --force %n.%u.%j.%s.%t"
  • nvidia-container-runtime using run (suggested): 
    EnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeEnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeQuery="nvidia-container-runtime --rootless=true --root=/run/user/%U/ state %n.%u.%j.%s.%t"
RunTimeKill="nvidia-container-runtime --rootless=true --root=/run/user/%U/ kill -a %n.%u.%j.%s.%t"
RunTimeDelete="nvidia-container-runtime --rootless=true --root=/run/user/%U/ delete --force %n.%u.%j.%s.%t"
RunTimeRun="nvidia-container-runtime --rootless=true --root=/run/user/%U/ run %n.%u.%j.%s.%t -b %b"
  • hpcng singularity v3.8.0: 
    EnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeEnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
OCIRunTimeQuery="sudo singularity oci state %n.%u.%j.%s.%t"
OCIRunTimeCreate="sudo singularity oci create --bundle %b %n.%u.%j.%s.%t"
OCIRunTimeStart="sudo singularity oci start %n.%u.%j.%s.%t"
OCIRunTimeKill="sudo singularity oci kill %n.%u.%j.%s.%t"
OCIRunTimeDelete="sudo singularity oci delete %n.%u.%j.%s.%t
    WARNING: Singuarity (v3.8.0) requires sudo for OCI support, which is a security risk since the user is able to modify these calls. This example is only provided for testing purposes.
  • Charliecloud (v0.30) 
    IgnoreFileConfigJson=true
CreateEnvFile=newline
EnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeEnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeRun="env -i PATH=/usr/local/bin:/usr/local/sbin:/usr/bin:/usr/sbin:/bin/:/sbin/ USER=$(whoami) HOME=/home/$(whoami)/ ch-run -w --bind /etc/group:/etc/group --bind /etc/passwd:/etc/passwd --bind /etc/slurm:/etc/slurm --bind %m:/var/run/slurm/ --bind /var/run/munge/:/var/run/munge/ --set-env=%e --no-passwd %r -- %@"
RunTimeKill="kill -s SIGTERM %p"
RunTimeDelete="kill -s SIGKILL %p"
  • Enroot (3.3.0)
    oci.conf: 
    IgnoreFileConfigJson=true
CreateEnvFile=newline
EnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeEnvExclude="^(SLURM_CONF|SLURM_CONF_SERVER)="
RunTimeRun="/usr/local/bin/enroot-start-wrapper %b %m %e -- %@"
RunTimeKill="kill -s SIGINT %p"
RunTimeDelete="kill -s SIGTERM %p"
    /usr/local/bin/enroot-start-wrapper: 
    #!/bin/bash
BUNDLE="$1"
SPOOLDIR="$2"
ENVFILE="$3"
shift 4
IMAGE=

export USER=$(whoami)
export HOME="$BUNDLE/"
export TERM
export ENROOT_SQUASH_OPTIONS='-comp gzip -noD'
export ENROOT_ALLOW_SUPERUSER=n
export ENROOT_MOUNT_HOME=y
export ENROOT_REMAP_ROOT=y
export ENROOT_ROOTFS_WRITABLE=y
export ENROOT_LOGIN_SHELL=n
export ENROOT_TRANSFER_RETRIES=2
export ENROOT_CACHE_PATH="$SPOOLDIR/"
export ENROOT_DATA_PATH="$SPOOLDIR/"
export ENROOT_TEMP_PATH="$SPOOLDIR/"
export ENROOT_ENVIRON="$ENVFILE"

if [ ! -f "$BUNDLE" ]
then
        IMAGE="$SPOOLDIR/container.sqsh"
        enroot import -o "$IMAGE" -- "$BUNDLE" && \
        enroot create "$IMAGE"
        CONTAINER="container"
else
        CONTAINER="$BUNDLE"
fi

enroot start -- "$CONTAINER" "$@"
rc=$?

[ $IMAGE ] && unlink $IMAGE

exit $rc

Testing OCI runtime outside of Slurm

Slurm calls the OCI runtime directly in the job step. If it fails, then the job will also fail.

  • Go to the directory containing the OCI Container bundle: 
    cd $ABS_PATH_TO_BUNDLE
  • Execute OCI Container runtime (You can find a few examples on how to build a bundle below): 
    $OCIRunTime $ARGS create test --bundle $PATH_TO_BUNDLE
    $OCIRunTime $ARGS start test
    $OCIRunTime $ARGS kill test
    $OCIRunTime $ARGS delete test
    If these commands succeed, then the OCI runtime is correctly configured and can be tested in Slurm.

Requesting container jobs or steps

salloc, srun and sbatch (in Slurm 21.08+) have the '--container' argument, which can be used to request container runtime execution. The requested job container will not be inherited by the steps called, excluding the batch and interactive steps.

  • Batch step inside of container: 
    sbatch --container $ABS_PATH_TO_BUNDLE --wrap 'bash -c "cat /etc/*rel*"'
  • Batch job with step 0 inside of container: 
    sbatch --wrap 'srun bash -c "--container $ABS_PATH_TO_BUNDLE cat /etc/*rel*"'
  • Interactive step inside of container: 
    salloc --container $ABS_PATH_TO_BUNDLE bash -c "cat /etc/*rel*"
  • Interactive job step 0 inside of container: 
    salloc srun --container $ABS_PATH_TO_BUNDLE bash -c "cat /etc/*rel*"
  • Job with step 0 inside of container: 
    srun --container $ABS_PATH_TO_BUNDLE bash -c "cat /etc/*rel*"
  • Job with step 1 inside of container: 
    srun srun --container $ABS_PATH_TO_BUNDLE bash -c "cat /etc/*rel*"

Integration with Rootless Docker (Docker Engine v20.10+ & Slurm-23.02+)

Slurm's scrun can be directly integrated with Rootless Docker to run containers as jobs. No special user permissions are required and should not be granted to use this functionality.

Prerequisites

  1. slurm.conf must be configured to use Munge authentication.
    AuthType=auth/munge
  2. scrun.lua must be configured for site storage configuration.

Limitations

  1. JWT authentication is not supported.
  2. Docker container building is not currently functional pending merge of Docker pull request.
  3. Docker does not expose configuration options to disable security options needed to run jobs. This requires that all calls to docker provide the following command line arguments. This can be done via shell variable, an alias, wrapper function, or wrapper script: 
    --security-opt label:disable --security-opt seccomp=unconfined --security-opt apparmor=unconfined --net=none
    Docker's builtin security functionality is not required (or wanted) for containers being run by Slurm. Docker is only acting as a container image lifecycle manager. The containers will be executed remotely via Slurm following the existing security configuration in Slurm outside of unprivileged user control.
  4. All containers must use host networking

Setup procedure

  1. Install and configure Rootless Docker
    Rootless Docker must be fully operational and able to run containers before continuing.
  2. Setup environment for all docker calls: 
    export DOCKER_HOST=unix://$XDG_RUNTIME_DIR/docker.sock
    All commands following this will expect this environment variable to be set.
  3. Stop rootless docker: 
    systemctl --user stop docker
  4. Configure Docker to call scrun instead of the default OCI runtime.
    • To configure for all users: 
      /etc/docker/daemon.json
    • To configure per user: 
      ~/.config/docker/daemon.json
    Set the following fields to configure Docker: 
    {
  "default-runtime": "slurm",
  "runtimes": {
    "slurm": {
      "path": "/usr/local/bin/scrun"
    }
  },
  "experimental": true,
  "iptables": false,
  "bridge": "none",
  "no-new-privileges": true,
  "rootless": true,
  "selinux-enabled": false,
}
    Correct path to scrun as if installation prefix was configured.
  5. It is strongly suggested that sites consider using inter-node shared filesystems to store Docker's containers. While it is possible to have a scrun.lua script to push and pull images for each deployment, there can be a massive performance penalty. Using a shared filesystem will avoid moving these files around.
    Possible configuration additions to daemon.json to use a shared filesystem with vfs storage driver: 
    {
  "storage-driver": "vfs",
  "data-root": "/path/to/shared/filesystem/user_name/data/",
  "exec-root": "/path/to/shared/filesystem/user_name/exec/",
}
    Any node expected to be able to run containers from Docker must have ability to atleast read the filesystem used. Full write privileges are suggested and will be required if changes to the container filesystem are desired.
  6. Start rootless docker: 
    systemctl --user start docker
  7. Verify Docker is using scrun: 
    export DOCKER_SECURITY="--security-opt label:disable --security-opt seccomp=unconfined  --security-opt apparmor=unconfined --net=none"
docker run $DOCKER_SECURITY hello-world
docker run $DOCKER_SECURITY alpine printenv SLURM_JOB_ID
docker run $DOCKER_SECURITY alpine hostname
docker run $DOCKER_SECURITY alpine -e SCRUN_JOB_NUM_NODES=10 hostname

Integration with Podman (Slurm-23.02+)

Slurm's scrun can be directly integrated with Podman to run containers as jobs. No special user permissions are required and should not be granted to use this functionality.

Prerequisites

  1. Slurm must be fully configured and running on host running dockerd.
  2. slurm.conf must be configured to use Munge authentication.
    AuthType=auth/munge
  3. scrun.lua must be configured for site storage configuration.

Limitations

  1. JWT authentication is not supported.
  2. All containers must use host networking

Setup procedure

  1. Install and configure Podman
    Podman must be fully operational and able to run containers before continuing.
  2. Configure Podman to call scrun instead of the default OCI runtime.
    • To configure for all users: 
      /etc/containers/containers.conf
    • To configure per user: 
      ~/.config/containers/containers.conf
    Set the following configuration parameters to configure Podman: 
    [containers]
apparmor_profile = "unconfined"
cgroupns = "host"
cgroups = "enabled"
default_sysctls = []
label = false
netns = "host"
no_hosts = true
pidns = "host"
utsns = "host"
userns = "host"

[engine]
runtime = "slurm"
runtime_supports_nocgroups = [ "slurm" ]
runtime_supports_json = [ "slurm" ]
remote = false

[engine.runtimes]
slurm = [ "/usr/local/bin/scrun" ]
    Correct path to scrun as if installation prefix was configured.
  3. It is strongly suggested that sites consider using inter-node shared filesystems to store Podman's containers. While it is possible to have a scrun.lua script to push and pull images for each deployment, there can be a massive performance penalty. Using a shared filesystem will avoid moving these files around.
    • To configure for all users: 
      /etc/containers/storage.conf
    • To configure per user: 
      ~/.config/containers/storage.conf
    Possible configuration additions to storage.conf to use a shared filesystem with vfs storage driver: 
    [storage]
driver = "vfs"
runroot = "$HOME/containers"
graphroot = "$HOME/containers"

[storage.options]
pull_options = {use_hard_links = "true", enable_partial_images = "true"}


[storage.options.vfs]
ignore_chown_errors = "true"
    Any node expected to be able to run containers from Podman must have ability to atleast read the filesystem used. Full write privileges are suggested and will be required if changes to the container filesystem are desired.
  4. Verify Podman is using scrun: 
    podman run hello-world
podman run alpine printenv SLURM_JOB_ID
podman run alpine hostname
podman run alpine -e SCRUN_JOB_NUM_NODES=10 hostname
salloc podman run --env-host=true alpine hostname
salloc sh -c 'podman run -e SLURM_JOB_ID=$SLURM_JOB_ID alpine hostname'

OCI Container bundle

There are multiple ways to generate an OCI Container bundle. The instructions below are the method we found the easiest. The OCI standard provides the requirements for any given bundle: Filesystem Bundle

Here are instructions on how to generate a container using a few alternative container solutions:

  • Create an image and prepare it for use with runc:
    1. Use an existing tool to create a filesystem image in /image/rootfs:
      • debootstrap: 
        sudo debootstrap stable /image/rootfs http://deb.debian.org/debian/
      • yum: 
        sudo yum --config /etc/yum.conf --installroot=/image/rootfs/ --nogpgcheck --releasever=${CENTOS_RELEASE} -y
      • docker: 
        mkdir -p ~/oci_images/alpine/rootfs
cd ~/oci_images/
docker pull alpine
docker create --name alpine alpine
docker export alpine | tar -C ~/oci_images/alpine/rootfs -xf -
docker rm alpine
    2. Configure a bundle for runtime to execute:
      • Use runc to generate a config.json: 
        cd ~/oci_images/alpine
runc --rootless=true spec --rootless
      • Test running image:
        • srun --container ~/oci_images/alpine/ uptime
  • Use umoci and skopeo to generate a full image: 
    mkdir -p ~/oci_images/
cd ~/oci_images/
skopeo copy docker://alpine:latest oci:alpine:latest
umoci unpack --rootless --image alpine ~/oci_images/alpine
srun --container ~/oci_images/alpine uptime
  • Use singularity to generate a full image: 
    mkdir -p ~/oci_images/alpine/
cd ~/oci_images/alpine/
singularity pull alpine
sudo singularity oci mount ~/oci_images/alpine/alpine_latest.sif ~/oci_images/alpine
mv config.json singularity_config.json
runc spec --rootless
srun --container ~/oci_images/alpine/ uptime

Example OpenMPI v5 + PMIx v4 container

Minimalist Dockerfile to generate a image with OpenMPI and PMIx to test basic MPI jobs.

Dockerfile

FROM almalinux:latest
RUN dnf -y update && dnf -y upgrade && dnf install -y yum-utils && dnf config-manager --set-enabled powertools
RUN dnf -y install make automake gcc gcc-c++ kernel-devel bzip2 python3 wget libevent-devel hwloc-devel munge-devel

WORKDIR /usr/local/src/
RUN wget 'https://github.com/openpmix/openpmix/releases/download/v4.2.2/pmix-4.2.2.tar.bz2' -O - | tar -xvjf -
WORKDIR /usr/local/src/pmix-4.2.2/
RUN ./configure && make -j && make install

WORKDIR /usr/local/src/
RUN wget --inet4-only 'https://download.open-mpi.org/release/open-mpi/v5.0/openmpi-5.0.0rc9.tar.gz' -O - | tar -xvzf -
WORKDIR /usr/local/src/openmpi-5.0.0rc9
RUN ./configure --disable-pty-support --enable-ipv6 --without-slurm --with-pmix --enable-debug && make -j && make install

WORKDIR /usr/local/src/openmpi-5.0.0rc9/examples
RUN make && cp -v hello_c ring_c connectivity_c spc_example /usr/local/bin

Container support via Plugin

Slurm also allows container developers to create SPANK Plugins that can be called at various points of job execution to support containers. Slurm is generally agnostic to SPANK based containers and can be made to start most, if not all, types. Any site using a plugin to start containers should not create or configure the "oci.conf" configuration file to deactivate the OCI container functionality.

Some container developers have chosen a command line interface only which requires users to explicitly execute the container solution.

Links to several third party container solutions are provided below:

Please note this list is not exhaustive as new containers types are being created all the time.

Container Types

Charliecloud

Charliecloud is user namespace container system sponsored by LANL to provide HPC containers. Charliecloud supports the following:

Docker (running as root)

Docker currently has multiple design points that make it unfriendly to HPC systems. The issue that usually stops most sites from using Docker is the requirement of "only trusted users should be allowed to control your Docker daemon" [Docker Security] which is not acceptable to most HPC systems.

Sites with trusted users can add them to the docker Unix group and allow them control Docker directly from inside of jobs. There is currently no direct support for starting or stopping docker containers in Slurm.

Sites are recommended to extract the container image from docker (procedure above) and then run the containers using Slurm.

UDOCKER

UDOCKER is Docker feature subset clone that is designed to allow execution of docker commands without increased user privileges.

Rootless Docker

Rootless Docker (>=v20.10) requires no extra permissions for users and currently (as of January 2021) has no known security issues with users gaining privileges. Each user will need to run an instance of the dockerd server on each node of the job in order to use docker. There are currently no helper scripts or plugins for Slurm to automate the build up or tear down the docker daemons.

Sites are recommended to extract the container image from docker (procedure above) and then run the containers using Slurm.

Kubernetes Pods (k8s)

Kubernetes is a container orchestration system that uses PODs, which are generally a logical grouping of containers for singular purpose.

There is currently no direct support for Kubernetes Pods in Slurm. Sites are encouraged to extract the OCI image from Kubernetes and then run the containers using Slurm. Users can create jobs that start together using the "--dependency=" argument in sbatch to mirror the functionality of Pods. Users can also use a larger allocation and then start each pod as a parallel step using srun.

Shifter

Shifter is a container project out of NERSC to provide HPC containers with full scheduler integration.

Singularity

Singularity is hybrid container system that supports:

  • Slurm integration (for singularity v2.x) via Plugin. A full description of the plugin was provided in the SLUG17 Singularity Presentation.
  • User namespace containers via sandbox mode that require no additional permissions.
  • Users directly calling singularity via setuid executable outside of Slurm.

ENROOT

Enroot is a user namespace container system sponsored by NVIDIA that supports:

  • Slurm integration via pyxis
  • Native support for Nvidia GPUs
  • Faster Docker image imports

Podman

Podman is a user namespace container system sponsored by Redhat/IBM that supports:

  • Drop in replacement of Docker.
  • Called directly by users. (Currently lacks direct Slurm support).
  • Rootless image building via buildah
  • Native OCI Image support

Sarus

Sarus is a privileged container system sponsored by ETH Zurich CSCS that supports:

Overview slides of Sarus are here.

Last modified 9 March 2023