Podman Guide #
This guide describes Podman and containers.
1 Introduction to containers #
Containers offer a lightweight virtualization method to run multiple virtual environments (containers) simultaneously on a single host. Unlike technologies such as Xen or KVM, where the processor simulates a complete hardware environment and a hypervisor controls virtual machines, containers provide virtualization on the operating system level, where the kernel controls the isolated containers.
Containers make it possible to isolate applications in self-contained units.
Containers provide near-native performance. Depending on the runtime, a container can use the host kernel directly, thus minimizing overhead.
It is possible to control network interfaces and apply resources inside containers through kernel control groups.
Containers run on the host system's kernel, so they cannot use different kernels or different kernel versions.
Containers security depends on the host system. Containerized applications can be secured through AppArmor or SELinux profiles. Securing containers is harder than securing virtual machines, due to the larger attack surface.
2 Podman overview #
Podman is a short name for Pod Manager Tool. It is a daemonless container engine that enables you to run and deploy applications using containers and container images. Podman provides a command line interface to manage containers.
As Podman does not have a daemon, it provides integration with systemd. This makes it possible to control containers via systemd units. You can create these units for existing containers as well as generate units that can start containers if they do not exist in the system. Moreover, Podman can run systemd inside containers.
Podman enables you to organize your containers into pods. Pods share the same network interface and resources. A typical use case for organizing a group of containers into a pod is a container that runs a database and a container with a client that accesses the database. For further information about pods, refer to Section 2.1, “Pods architecture”.
2.1 Pods architecture #
A pod is a group of containers that share the same name space, ports,
and network connection. Usually, containers within one pod can
communicate directly with each other. Each pod contains an
infrastructure container (INFRA
), whose purpose is
to hold the name space. INFRA
also enables Podman to
add other containers to the pod. Port bindings, cgroup-parent values,
and kernel name spaces are all assigned to the infrastructure
container. Therefore, later changes of these values are not possible.
Each container in a pod has its own instance of a monitoring program
(conmon
). The monitoring program watches the
container's process and if the container dies, the monitoring program
saves its exit code. The program also holds open the tty interface for
the particular container. The monitoring program enables you to run
containers in the detached mode when Podman exits, because this program
continues to run and enables you to attach tty later.
3 Getting Podman #
In case of SLE Micro, Podman is delivered in all raw images or is installed by default if you installed your system manually from ISO. To verify that Podman is installed in your system, run the following command:
#
zypper se -i podman
If Podman is not listed in the output, install it by running:
#
transactional-update pkg install podman
In the Cockpit web interface, you need to start Podman as a service by clicking
to access container management.
By default, Podman requires root
privileges. To enable rootless
mode for the current user, run the following command:
>
sudo usermod --add-subuids 100000-165535 \
--add-subgids 100000-165535 USER
Reboot the machine to enable the change. The command above defines a
range of local UIDs to which the UIDs allocated to users inside the
container are mapped on the host. Note that the ranges defined for
different users must not overlap. It is also important that the ranges do
not reuse the UID of an existing local user or group. By default, adding
a user with the useradd
command automatically
allocates subUID and subGID ranges.
Running a container with Podman in rootless mode on SLE Micro may fail,
because the container might need access to directories or files that
require root
privileges.
The toolbox container also requires root
privileges.
4 Getting container images #
To run a container, you need an image. An image includes all dependencies
needed to run an application. You can obtain images from an image
registry. Available registries are defined in the
/etc/containers/registries.conf
configuration file.
If you have a local image registry or want to use other registries, add
the registries into the configuration file.
SLE Micro does not provide tools for building custom images. Therefore, the only way to get an image is to pull it from an image registry.
The podman pull
command pulls an image from an image
registry. The syntax is as follows:
#
podman pull [OPTIONS] SOURCE
The source can be an image without the
registry name. In that case, Podman tries to pull the image from all
registries configured in the
/etc/containers/registries.conf
file. The default
image tag is latest
. The default location of pulled
images is
/var/lib/containers/storage/overlay-images/
.
To view all possible options of the podman pull
command, run:
#
podman pull --help
If you are using Cockpit, you can also pull images from an image registry in the
menu by clicking .Podman enables you to search for images in an image registry or a list of registries using the command:
#
podman search IMAGE_NAME
Alternatively, you can use the skopeo tool to manage container images and image repositories. For details, refer to the following section.
4.1 skopeo #
skopeo is a command-line utility for managing, inspecting and signing container images and image repositories. skopeo can be used to inspect containers and repositories on remote and local container registries. skopeo can also copy container images between different storage back-ends.
skopeo works with the following registry types:
- containers-storage:IMAGE_REFERENCE
An image located in a local image store.
- docker://IMAGE_REFERENCE
An image in a registry.
skopeo provides several commands to manage images and registries:
inspect
This command fetches the repository manifest and can show you information like tags available for the specified repository, labels of a container image, the operating system of an image, and so on.
The command has the following syntax:
>
skopeo inspect REGISTRY_TYPEIMAGE_NAME
Here is a usage example:
>
skopeo inspect docker://registry.suse.com/suse/pcp:latest
{ "Name": "registry.suse.com/suse/pcp", "Digest": "sha256:eee17c009fb8b05e5825a8c9658d972ab13a17541180bd7a1348fccc6e4fc77f", "RepoTags": [ "5", "5-12.54", "5-13.10", ... ], "Created": "2023-06-19T16:59:01.617731565Z", "DockerVersion": "20.10.23-ce", "Labels": { "com.suse.application.pcp.created": "2023-06-19T16:58:29.786850402Z", "com.suse.application.pcp.description": "Performance Co-Pilot (pcp) container image based on the SLE Base Container Image. This container image is not supported when using a container runtime other than podman.", ... ], "Architecture": "amd64", "Os": "linux", "Layers": [ "sha256:a05e4c4d1fc89c7f7aa60829b8631b64873df6041f627eb2b43524e9e010446e", "sha256:19793da49ce886a67aa62657dc24a105a26d63568ce27de241246bc6cc9bc008", "sha256:403f9ef6c98d4cf277caf3166ca4455817828e33c8b699237bb8eb24cb2b41bf" ], "LayersData": [ { "MIMEType": "application/vnd.docker.image.rootfs.diff.tar.gzip", "Digest": "sha256:a05e4c4d1fc89c7f7aa60829b8631b64873df6041f627eb2b43524e9e010446e", "Size": 47291175, "Annotations": null }, ... ], "Env": [ "PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin"copy
This command enables you to copy container images between registries, container storage back-ends and local directories.
Here is a usage example:
>
skopeo copy oci:busybox_ocilayout:latest dir:existingemptydirectory
delete
To mark an image for later removal by the registry's garbage collector.
>
skopeo delete docker://registry.example.com/example/pause:latest
sync
To synchronize images between registry repositories and local directories.
5 Working with containers #
The following section covers common container management tasks. This includes creating, starting, and modifying containers.
5.1 Running containers #
After you have pulled your container image, you can create containers
based on it. You can run an instance of the image using the
podman run
command. The command syntax is as
follows:
#
podman run [OPTIONS] IMAGE [CONTAINER_NAME]
IMAGE is specified in format
transport:path. If transport
is omitted, the default docker
is used. The
path can reference to a specific image registry.
If omitted, Podman searches for the image in registries defined in the
/etc/containers/registries.conf
file. An example
that runs a container called sles15
based on the
sle15
image follows:
#
podman run registry.opensuse.org/suse/templates/images/sle-15-sp4/base/images/suse/sle15 sles15
Below is a list of frequently used options. For a complete list of
available options, run the command: podman run
--help
.
--detach, -d
The container will run in the background.
--env, -e=env
This option allows arbitrary environment variables that are available for the process to be launched inside of the container. If an environment variable is specified without a value, Podman will check the host environment for a value and set the variable only if it is set on the host.
--help
Prints help for the
podman run
command.--hostname=
name,-h
Sets the container host name that is available inside the container.
--pod=
nameRuns container in an existing pod. To create a pod, prefix the pod name with
new:
.--read-only
Mounts the container’s root file system as read only.
--systemd=true|false|always
Runs container in systemd mode. The default is true.
5.2 Stopping containers #
If the podman run
command finished successfully, a
new container has been started. You can stop the container by running:
#
podman stop [OPTIONS] CONTAINER
You can specify a single container name or ID or a space-separated list of containers. The command takes the following options:
--all, -a
Stops all running containers.
--latest, -l
Instead of providing a container name, the last created container will be stopped.
--time, -t=
secondsSeconds to wait before forcibly stopping the container.
To view all possible options of the podman stop
command, run the following:
#
podman stop --help
5.3 Starting containers #
To start already created but stopped containers, use the
podman start
command. The command syntax is as
follows:
#
podman start [OPTIONS] CONTAINER
CONTAINER can be a container name or a container ID.
For a complete list of possible options of podman
start
, run the command:
#
podman start --help
5.4 Committing modified containers #
You can run a new container with specific attributes that are not part
of the original image. To save the container with these attributes as a
new image, you can use the podman commit
command:
#
podman commit [OPTIONS] CONTAINER IMAGE
CONTAINER is a container name or a container
ID. IMAGE is the new image name. If the
image name does not start with a registry name, the value
localhost
is used.
5.5 Listing containers #
Podman enables you to list all running containers using the
podman ps
command. The generic syntax of the command
is as follows:
#
podman ps [OPTIONS]
Command options can change the displayed information. For example,
using the --all
option will output all containers
created by Podman (not only the running containers).
For a complete list of podman ps
options, run:
#
podman ps --help
5.6 Removing containers #
To remove one or more unused container from the host, use the
podman rm
command as follows:
#
podman rm [OPTIONS] CONTAINER
CONTAINER can be a container name or a container ID.
The command does not remove the specified container if the container is
running. To remove a running container, use the -f
option.
For a complete list of podman rm
options, run:
#
podman rm --help
You can delete all stopped containers from your host with a single command:
#
podman container prune
Make sure that each stopped container is intended to be removed before you run the command, otherwise you might remove containers that are still in use and were stopped only temporarily.
6 Working with pods #
Containers can be grouped into a pod. The containers in the pod then
share network, pid, and IPC namespace. Pods can be managed by
podman pod
commands. This section provides an overview
of the commands for managing pods.
6.1 Creating pods #
The command podman pod create
is used to create a
pod. The syntax of the command is as follows:
#
podman pod create [OPTIONS]
The command outputs the pod ID. By default, the pods are created without being started. You can start a pod by running a container in the pod, or by starting the pod as described in Section 6.3, “Starting/stopping/restarting pods”.
If you do not specify a pod name with the --name
option, Podman will assign a default name for the pod.
For a complete list of possible options, run the following command:
#
podman pod create --help
6.2 Listing pods #
You can list all pods by running the command:
#
podman pod list
The output looks as follows:
POD ID NAME STATUS CREATED # OF CONTAINERS INFRA ID 30fba506fecb upbeat_mcclintock Created 19 hours ago 1 4324f40c9651 976a83b4d88b nervous_feynman Running 19 hours ago 2 daa5732ecd02
As each pod includes the INFRA
container, the number
of containers in a pod is always larger than zero.
6.3 Starting/stopping/restarting pods #
After a pod is created, you must start it, as it is not in the state
running
by default. In the commands below,
POD can be a pod name or a pod ID.
To start a pod, run the command:
#
podman pod start [OPTIONS] POD
For a complete list of possible options, run:
#
podman pod start --help
To stop a pod, use the podman pod stop
as follows:
#
podman pod stop POD
To restart a pod, use the podman pod restart
command
as follows:
#
podman pod restart POD
6.4 Managing containers in a pod #
To add a new container to a pod, use the podman run
command with the option --pod
. A general syntax of
the command follows:
#
podman run [OPTIONS] --pod POD_NAME IMAGE
For details about the podman run
command, refer to
Section 5.1, “Running containers”.
The podman start
command does not allow for
starting a container in a pod if the container was not added to the
pod during the container's initial running.
You cannot remove a container from a pod and keep the container running, because the container itself is removed from the host.
Other actions like start, restart, and stop can be performed on specific containers without affecting the status of the pod.
6.5 Removing pods #
There are two ways to remove pods. You can use the podman pod
rm
command to remove one or more pods. Alternatively, you can
remove all stopped pods using the podman pod prune
command.
To remove a pod or several pods, run the podman pod
rm
command as follows:
#
podman pod rm POD
POD can be a pod name or a pod ID.
To remove all currently stopped pods, use the podman pod
prune
command. Make sure that all stopped pods are intended
to be removed before you run the podman pod prune
command, otherwise you might remove pods that are still in use.
6.6 Monitoring processes in pods #
To view all containers in all pods, use the following command:
#
podman ps -a --pod
The output of the command will be similar to the following one:
CONTAINER ID IMAGE COMMAND CREATED STATUS [...] 4324f40c9651 k8s.gcr.io/pause:3.2 21 hours ago Created daa5732ecd02 k8s.gcr.io/pause:3.2 22 hours ago Up 3 hours ago e5c8e360c54b localhost/test:latest /bin/bash 3 days ago Exited (137) 3 days ago 82dad15828f7 localhost/opensuse/toolbox /bin/bash 3 days ago Exited (137) 3 days ago 1a23da456b6f docker.io/i386/ubuntu /bin/bash 4 days ago Exited (0) 6 hours ago df890193f651 localhost/opensuse/toolbox /bin/bash 4 days ago Created
The first two records are the INFRA
containers of
each pod, based on the k8s.gcr.io/pause:3.2
image.
Other containers in the output are stand-alone containers that do not
belong to any pod.
Terminology
- Container
A container is a running instance based on a particular container image. Each container can be distinguished by a unique container ID.
- Control groups
Control groups, also called
cgroups
, are a Linux kernel feature that allows aggregating or partitioning tasks (processes) and all their children into hierarchically-organized groups, to manage their resource limits.- Container image
A container image is an unchangeable, static file that includes executable code so it can run an isolated process on IT infrastructure. The image is comprised of system libraries, system tools, and other platform settings a program needs to run on a containerization platform. A container image is compiled from file system layers built on top of a parent or base image.
- Base image
A base image is an image that does not have a parent image. In a Dockerfile, a base image is identified by the
FROM scratch
directive.- Parent image
The image that serves as the basis for another container image. In other words, if an image is not a base image, it is derived from a parent image.
- Namespaces
Docker Open Source Engine uses Linux namespaces for its containers, which isolates resources reserved for particular containers.
- Orchestration
In a production environment, you typically need a cluster with many containers on each cluster node. The containers must cooperate and you need a framework that enables you to automatically manage the containers. The act of automatic container management is called container orchestration and is typically handled by Kubernetes.
- Registry
A registry is storage for already-created images. It typically contains several repositories. There are two types of registries:
public registry: Any (usually registered) user can download and use images. A typical example of a public registry is Docker Hub.
private registry: Access is restricted to particular users, or from a particular private network.
- Repository
A repository is storage for images in a registry.