ovirt-imageio/container/Containerfile

# SPDX-FileCopyrightText: 2022 Red Hat, Inc.
# SPDX-License-Identifier: GPL-2.0-or-later

# Build stage

FROM alpine AS build
RUN apk add --no-cache --update build-base linux-headers python3 python3-dev py3-pip
WORKDIR /build
COPY ovirt-imageio.tar.gz .
RUN python3 -m venv /venv
RUN /venv/bin/pip install ovirt-imageio.tar.gz

# Final stage

FROM alpine
RUN apk add python3 qemu-img
RUN mkdir /etc/ovirt-imageio
COPY --from=build /venv /venv
COPY conf.d /etc/ovirt-imageio/conf.d
COPY app /app
EXPOSE 80/tcp
ENTRYPOINT [ "/app/entrypoint.py" ]
reuse: addheader container folder Add SPDX header to files in the container directory. Signed-off-by: Albert Esteve <aesteve@redhat.com> 2022-10-17 14:32:15 +02:00			`# SPDX-FileCopyrightText: 2022 Red Hat, Inc.`
			`# SPDX-License-Identifier: GPL-2.0-or-later`

container: Add ovirt-imageio container Start a container image for running imageio and container. Using a container, the user will run an image transfer pod for every transfer, exposing the pod for remote access, and connecting the relevant storage the pod. When the transfer is done, the user will unexpose the pod and terminate it. In this mode the imageio server will serve one ticket, injected into the pod when starting the pod, instead of adding a ticket using the control service. To created the container run: make container This creates localhost/ovirt-imageio container image using podman configured for rootless environment[1]. To run the container: podman run --publish 8080:80 -v /path/to/image:/images:Z ovirt-imageio Next we only need to load a ticket and create the image in the mounted images folder. Note that the path for the file in the ticket needs to match the path where we mounted the images folder in the container. Alternatively, we can mount the nbd socket path in the container and request the image with a nbd ticket. More work is needed: - Modify the server to load a ticket from file during startup - Document how to mount the image to transfer or the nbd socket in the container. - Inject PKI files into the container (can be done later) [1] https://github.com/containers/podman/blob/main/docs/tutorials/rootless_tutorial.md Signed-off-by: Nir Soffer <nsoffer@redhat.com> 2022-06-12 22:26:12 +03:00			`# Build stage`

			`FROM alpine AS build`
			`RUN apk add --no-cache --update build-base linux-headers python3 python3-dev py3-pip`
			`WORKDIR /build`
			`COPY ovirt-imageio.tar.gz .`
			`RUN python3 -m venv /venv`
			`RUN /venv/bin/pip install ovirt-imageio.tar.gz`

			`# Final stage`

			`FROM alpine`
container: serve image with nbd Include qemu-nbd in the container to make it easier for users to use the containerized imageio server. The container now runs with a script as an ENTRYPOINT. The script runs as a program that can receive additional command line parameters. The script is basically a launcher for the two main services involved: - qemu-nbd serving the image - ovirt-imageio exposing the image through http This commit removes support for file ticket, only nbd tickets are implicitely used. This is better, as nbd does much of the hard work for us, so will perform better. Users will not have to worry about the ticket anyway, the ticket is also generated automatically, so no need to mount the ticket inside the container, or give an environment variable. Users can set the ticket ID in the command line, or use the file name as default. Example session: $ mkdir /var/tmp/images $ qemu-img create -f raw /var/tmp/images/disk.raw 6g $ podman run \ --interactive \ --tty \ --rm \ --publish 8080:80 \ --volume /var/tmp/images:/var/tmp:Z \ localhost/ovirt-imageio:latest /var/tmp/disk.raw 2022-10-20 14:46:08,953 INFO (MainThread) [server] Starting (hostname=1594cc8a121b pid=8, version=2.4.7) 2022-10-20 14:46:08,970 INFO (MainThread) [services] remote.service listening on ('::', 80) 2022-10-20 14:46:08,970 INFO (MainThread) [server] Initial ticket: /ticket.json 2022-10-20 14:46:08,971 INFO (MainThread) [server] Ready for requests ... 2022-10-26 12:07:00,155 INFO (MainThread) [server] Received signal 2, shutting down Interrupted, shutting down $ examples/imageio-client download -f raw http://localhost:8080/images/disk.raw download.raw [ 100% ] 6.00 GiB, 0.10 s, 57.23 GiB/s Signed-off-by: Albert Esteve <aesteve@redhat.com> 2022-10-20 16:48:40 +02:00			`RUN apk add python3 qemu-img`
container: Add ovirt-imageio container Start a container image for running imageio and container. Using a container, the user will run an image transfer pod for every transfer, exposing the pod for remote access, and connecting the relevant storage the pod. When the transfer is done, the user will unexpose the pod and terminate it. In this mode the imageio server will serve one ticket, injected into the pod when starting the pod, instead of adding a ticket using the control service. To created the container run: make container This creates localhost/ovirt-imageio container image using podman configured for rootless environment[1]. To run the container: podman run --publish 8080:80 -v /path/to/image:/images:Z ovirt-imageio Next we only need to load a ticket and create the image in the mounted images folder. Note that the path for the file in the ticket needs to match the path where we mounted the images folder in the container. Alternatively, we can mount the nbd socket path in the container and request the image with a nbd ticket. More work is needed: - Modify the server to load a ticket from file during startup - Document how to mount the image to transfer or the nbd socket in the container. - Inject PKI files into the container (can be done later) [1] https://github.com/containers/podman/blob/main/docs/tutorials/rootless_tutorial.md Signed-off-by: Nir Soffer <nsoffer@redhat.com> 2022-06-12 22:26:12 +03:00			`RUN mkdir /etc/ovirt-imageio`
			`COPY --from=build /venv /venv`
			`COPY conf.d /etc/ovirt-imageio/conf.d`
container: serve image with nbd Include qemu-nbd in the container to make it easier for users to use the containerized imageio server. The container now runs with a script as an ENTRYPOINT. The script runs as a program that can receive additional command line parameters. The script is basically a launcher for the two main services involved: - qemu-nbd serving the image - ovirt-imageio exposing the image through http This commit removes support for file ticket, only nbd tickets are implicitely used. This is better, as nbd does much of the hard work for us, so will perform better. Users will not have to worry about the ticket anyway, the ticket is also generated automatically, so no need to mount the ticket inside the container, or give an environment variable. Users can set the ticket ID in the command line, or use the file name as default. Example session: $ mkdir /var/tmp/images $ qemu-img create -f raw /var/tmp/images/disk.raw 6g $ podman run \ --interactive \ --tty \ --rm \ --publish 8080:80 \ --volume /var/tmp/images:/var/tmp:Z \ localhost/ovirt-imageio:latest /var/tmp/disk.raw 2022-10-20 14:46:08,953 INFO (MainThread) [server] Starting (hostname=1594cc8a121b pid=8, version=2.4.7) 2022-10-20 14:46:08,970 INFO (MainThread) [services] remote.service listening on ('::', 80) 2022-10-20 14:46:08,970 INFO (MainThread) [server] Initial ticket: /ticket.json 2022-10-20 14:46:08,971 INFO (MainThread) [server] Ready for requests ... 2022-10-26 12:07:00,155 INFO (MainThread) [server] Received signal 2, shutting down Interrupted, shutting down $ examples/imageio-client download -f raw http://localhost:8080/images/disk.raw download.raw [ 100% ] 6.00 GiB, 0.10 s, 57.23 GiB/s Signed-off-by: Albert Esteve <aesteve@redhat.com> 2022-10-20 16:48:40 +02:00			`COPY app /app`
container: Add ovirt-imageio container Start a container image for running imageio and container. Using a container, the user will run an image transfer pod for every transfer, exposing the pod for remote access, and connecting the relevant storage the pod. When the transfer is done, the user will unexpose the pod and terminate it. In this mode the imageio server will serve one ticket, injected into the pod when starting the pod, instead of adding a ticket using the control service. To created the container run: make container This creates localhost/ovirt-imageio container image using podman configured for rootless environment[1]. To run the container: podman run --publish 8080:80 -v /path/to/image:/images:Z ovirt-imageio Next we only need to load a ticket and create the image in the mounted images folder. Note that the path for the file in the ticket needs to match the path where we mounted the images folder in the container. Alternatively, we can mount the nbd socket path in the container and request the image with a nbd ticket. More work is needed: - Modify the server to load a ticket from file during startup - Document how to mount the image to transfer or the nbd socket in the container. - Inject PKI files into the container (can be done later) [1] https://github.com/containers/podman/blob/main/docs/tutorials/rootless_tutorial.md Signed-off-by: Nir Soffer <nsoffer@redhat.com> 2022-06-12 22:26:12 +03:00			`EXPOSE 80/tcp`
container: serve image with nbd Include qemu-nbd in the container to make it easier for users to use the containerized imageio server. The container now runs with a script as an ENTRYPOINT. The script runs as a program that can receive additional command line parameters. The script is basically a launcher for the two main services involved: - qemu-nbd serving the image - ovirt-imageio exposing the image through http This commit removes support for file ticket, only nbd tickets are implicitely used. This is better, as nbd does much of the hard work for us, so will perform better. Users will not have to worry about the ticket anyway, the ticket is also generated automatically, so no need to mount the ticket inside the container, or give an environment variable. Users can set the ticket ID in the command line, or use the file name as default. Example session: $ mkdir /var/tmp/images $ qemu-img create -f raw /var/tmp/images/disk.raw 6g $ podman run \ --interactive \ --tty \ --rm \ --publish 8080:80 \ --volume /var/tmp/images:/var/tmp:Z \ localhost/ovirt-imageio:latest /var/tmp/disk.raw 2022-10-20 14:46:08,953 INFO (MainThread) [server] Starting (hostname=1594cc8a121b pid=8, version=2.4.7) 2022-10-20 14:46:08,970 INFO (MainThread) [services] remote.service listening on ('::', 80) 2022-10-20 14:46:08,970 INFO (MainThread) [server] Initial ticket: /ticket.json 2022-10-20 14:46:08,971 INFO (MainThread) [server] Ready for requests ... 2022-10-26 12:07:00,155 INFO (MainThread) [server] Received signal 2, shutting down Interrupted, shutting down $ examples/imageio-client download -f raw http://localhost:8080/images/disk.raw download.raw [ 100% ] 6.00 GiB, 0.10 s, 57.23 GiB/s Signed-off-by: Albert Esteve <aesteve@redhat.com> 2022-10-20 16:48:40 +02:00			`ENTRYPOINT [ "/app/entrypoint.py" ]`