DocsReleasesCommunityGuidesBlog

Behind the Scenes with the Application Catalog

This guide presents internal technical information about the application catalog. It shows what is happening behind the scenes and how you can get more control on the build and run phases.

The guide on using the application catalog provides the user-friendly information on using the Unikraft application registry and the catalog repository. It presents some hints into what's happening behind the scenes, but it aims to keep the user away for this. This guide takes a deep dive into the internals of configuring, building and running Unikernel applications from the catalog. It is aimed for those more technically inclined who would be interested in understanding what's happening behind the scenes and maybe contribute to the application catalog.

Similar to the guide on using the application catalog, we will use two applications:

NGINX#

For the nginx/1.25 bincompat application, there is a build phase and a run phase. The build phase creates the output kernel, and the run phase launches a Unikraft virtual machine instance from the kernel.

The kernel is a join of the actual Unikraft kernel and the application filesystem, packed as an initial ramdisk. We call the packed initial ramdisk the embedded initial ramdisk or embedded initrd.

Configuration#

The build and run configuration is part of the Kraftfile.

The Kraftfile defines the:

  • resulting image name: nginx
  • the command line to start the application: /usr/bin/nginx
  • path to the template app-elfloader
  • paths and versions of repositories (unikraft, lwip, libelf)
  • configuration options: i.e. the CONFIG_... option enables the emdedded initrd build
  • build and run targets: currently only x86_64-based builds are available, and only KVM-based builds, using QEMU or Firecracker
  • root filesystem used to build the (embedded) initrd

The root filesystem is generated from a Dockerfile specification, as configured in the Kraftfile. The Dockerfile specification collects the required files (binary executable, depending libraries, configuration files, data files):

FROM --platform=linux/x86_64 nginx:1.25.3-bookworm AS build


# These are normally syminks to /dev/stdout and /dev/stderr, which don't
# (currently) work with Unikraft. We remove them, such that NGINX will create
# them by hand.
RUN rm /var/log/nginx/error.log
RUN rm /var/log/nginx/access.log


FROM scratch


# NGINX binaries, modules, configuration, log and runtime files
COPY --from=build /usr/sbin/nginx /usr/bin/nginx
COPY --from=build /usr/lib/nginx /usr/lib/nginx
COPY --from=build /etc/nginx /etc/nginx
COPY --from=build /etc/passwd /etc/passwd
COPY --from=build /etc/group /etc/group
COPY --from=build /var/log/nginx /var/log/nginx
COPY --from=build /var/cache/nginx /var/cache/nginx
COPY --from=build /var/run /var/run


# Libraries
COPY --from=build /lib/x86_64-linux-gnu/libcrypt.so.1 /lib/x86_64-linux-gnu/libcrypt.so.1
COPY --from=build /lib/x86_64-linux-gnu/libpcre2-8.so.0 /lib/x86_64-linux-gnu/libpcre2-8.so.0
COPY --from=build /lib/x86_64-linux-gnu/libssl.so.3 /lib/x86_64-linux-gnu/libssl.so.3
COPY --from=build /lib/x86_64-linux-gnu/libcrypto.so.3 /lib/x86_64-linux-gnu/libcrypto.so.3
COPY --from=build /lib/x86_64-linux-gnu/libz.so.1 /lib/x86_64-linux-gnu/libz.so.1
COPY --from=build /lib/x86_64-linux-gnu/libc.so.6 /lib/x86_64-linux-gnu/libc.so.6
COPY --from=build /lib64/ld-linux-x86-64.so.2 /lib64/ld-linux-x86-64.so.2
COPY --from=build /etc/ld.so.cache /etc/ld.so.cache


# Custom configuration files, including using a single process for Nginx
COPY ./conf/nginx.conf /etc/nginx/nginx.conf
COPY ./conf/unikraft.local.crt /etc/nginx/unikraft.local.crt
COPY ./conf/unikraft.local.key /etc/nginx/unikraft.local.key


# Web root
COPY ./wwwroot /wwwroot

The Dockerfile is being interpreted via BuildKit, hence the need to set up the BuildKit container.

Build Phase#

The build command requires the BuildKit container to be configured beforehand:

docker run -d --name buildkitd --privileged moby/buildkit:latest
export KRAFTKIT_BUILDKIT_HOST=docker-container://buildkitd

The build command is:

kraft build --plat qemu --arch x86_64

kraft build goes through the following steps:

  1. It generates the root filesystem, via BuildKit from the Dockerfile specification.
  2. It packs the root filsystem in an initial ramdisk (initrd).
  3. It builds the kernel, using the configuration in the Kraftfile.
  4. It embeds the initrd in the output kernel file.

The resulting embedded kernel image is .unikraft/build/nginx_qemu-x86_64:

$ ls -lh .unikraft/build/nginx_qemu-x86_64
-rwxr-xr-x 2 razvand docker 15M Jan  2 21:23 .unikraft/build/nginx_qemu-x86_64

Run Phase#

This image is run with a command such as:

kraft run -W -p 8080:80 .

It can also be run manually with qemu-system-x86_64:

qemu-system-x86_64 \
    -kernel .unikraft/build/nginx_qemu-x86_64 \
    -nographic \
    -m 128M \
    -device virtio-net-pci,mac=02:b0:b0:d3:d2:01,netdev=hostnet0 \
    -netdev user,id=hostnet0,hostfwd=tcp::8080-:80 \
    -append "/usr/bin/nginx" \
    -cpu max

This starts a QEMU virtual machine instance. Query it using:

curl http://localhost:8080

If you want use a bridge interface, first create the bridge interface as root (prefix with sudo if required):

kraft net create -n 172.44.0.1/24 virbr0

An the run manually with qemu-system-x86_64 as root (prefix with sudo if required):

qemu-system-x86_64 \
    -kernel .unikraft/build/nginx_qemu-x86_64 \
    -nographic \
    -m 128M \
    -netdev bridge,id=en0,br=virbr0 -device virtio-net-pci,netdev=en0 \
    -append "netdev.ip=172.44.0.2/24:172.44.0.1 -- /usr/bin/nginx" \
    -cpu max

This starts a QEMU virtual machine instance. Query it using:

curl http://172.44.0.2

To close the running QEMU instance, use Ctrl+a x in the QEMU console.

HTTP Go Server#

For the http-go1.21 bincompat example, there is no build phase, only a run phase. The example it's using a prebuilt kernel image. The prebuilt base kernel image is pulled from the registry, from unikraft.org/base. This happens during the run phase.

Configuration#

The run configuration is part of the Kraftfile:

spec: v0.6


runtime: base:latest


rootfs: ./Dockerfile


cmd: ["/server"]

The Kraftfile defines:

  • the runtime image to use, containing the kernel: unikraft.org/base:latest' (it can be summarized as just base:latest`)
  • the root filesystem used, defined in a Dockerfile
  • the command line to start the application: /server
  • the available run targets: currently only x86_64-based builds are available, and only KVM-based builds, using QEMU or Firecracker

The root filesystem is generated from a Dockerfile specification, as configured in the Kraftfile. The Dockerfile specification collects the required files (binary executable, depending libraries, configuration files, data files):

FROM golang:1.21.3-bookworm AS build


WORKDIR /src


COPY ./server.go /src/server.go


RUN set -xe; \
    CGO_ENABLED=1 \
    go build \
      -buildmode=pie \
      -ldflags "-linkmode external -extldflags '-static-pie'" \
      -tags netgo \
      -o /server server.go \
    ;


FROM scratch


COPY --from=build /server /server
COPY --from=build /lib/x86_64-linux-gnu/libc.so.6 /lib/x86_64-linux-gnu/
COPY --from=build /lib64/ld-linux-x86-64.so.2 /lib64/

The Dockerfile is being interpreted via BuildKit, hence the need to set up the BuildKit container.

Run Phase#

The run command requires the BuildKit container to be configured beforehand:

docker run -d --name buildkitd --privileged moby/buildkit:latest
export KRAFTKIT_BUILDKIT_HOST=docker-container://buildkitd

The run command is:

kraft run -W -p 8080:8080 .

kraft run goes through the following steps:

  1. It pulls the kernel package from the registry, from unikraft.org/base:latest.
  2. It generates the root filesystem, via BuildKit from the Dockerfile specification. The generation of the root filesystem implies the building the Go source code files into a binary executable (ELF). The executable, together with the depending libraries is then extracted into the root filesystem.
  3. It packs the root filesystem in an initial ramdisk (initrd).
  4. It runs the kernel attaching the initrd and using the command line in the specification: /http_server.

The resulting initrd image is .unikraft/build/initramfs.cpio.

$ ls -lh .unikraft/build/initramfs.cpio
-rw-r--r-- 1 root root 8.9M Jan  4 18:16 .unikraft/build/initramfs-x86_64.cpio

To view the contents of the root filesystem you can use cpio:

$ cpio -itv < .unikraft/build/initramfs.cpio
d---------   0 root     root            0 Jan  1  1970 /lib
d---------   0 root     root            0 Jan  1  1970 /lib/x86_64-linux-gnu
-rwxr-xr-x   1 root     root      1922136 Sep 30 11:31 /lib/x86_64-linux-gnu/libc.so.6
d---------   0 root     root            0 Jan  1  1970 /lib64
-rwxr-xr-x   1 root     root       210968 Sep 30 11:31 /lib64/ld-linux-x86-64.so.2
-rwxr-xr-x   1 root     root      7151306 Jan  4 18:16 /server
18136 blocks

The kernel image is pulled into a temporary directory.

To run the application manually, first pull the kernel image from unikraft.org/base:latest:

kraft pkg pull -w base unikraft.org/base:latest

The kernel image is base/unikraft/bin/kernel:

$ tree
base/
`-- unikraft/
    `-- bin/
        `-- kernel


3 directories, 1 file


$ ls -lh base/unikraft/bin/kernel
-rw-rw-r-- 1 razvand razvand 1.6M Jan 25 14:48 base/unikraft/bin/kernel

You can run the application manually with qemu-system-x86_64 and the passing of the -kernel, -initrd and -append arguments:

qemu-system-x86_64 \
    -kernel base/unikraft/bin/kernel \
    -nographic \
    -m 128M \
    -device virtio-net-pci,mac=02:b0:b0:d3:d2:01,netdev=hostnet0 \
    -netdev user,id=hostnet0,hostfwd=tcp::8080-:8080 \
    -append "vfs.fstab=[ \"initrd0:/:extract:::\" ] -- /server" \
    -initrd .unikraft/build/initramfs-x86_64.cpio \
    -cpu max

This starts a QEMU virtual machine instance. Query it using:

curl http://localhost:8080

If you want use a bridge interface, first create the bridge interface as root (prefix with sudo if required):

kraft net create -n 172.44.0.1/24 virbr0

An the run manually with qemu-system-x86_64 as root (prefix with sudo if required):

qemu-system-x86_64 \
    -kernel base/unikraft/bin/kernel \
    -nographic \
    -m 128M \
    -netdev bridge,id=en0,br=virbr0 -device virtio-net-pci,netdev=en0 \
    -append "netdev.ip=172.44.0.2/24:172.44.0.1 vfs.fstab=[ \"initrd0:/:extract:::\" ] -- /server" \
    -initrd .unikraft/build/initramfs-x86_64.cpio \
    -cpu max

This starts a QEMU virtual machine instance. Query it using:

curl http://172.44.0.2:8080

To close the running QEMU instance, use Ctrl+a x in the QEMU console.

Edit this page on GitHub

Connect with the community

Feel free to ask questions, report issues, and meet new people.

Join us on Discord!
®

Getting Started

What is a unikernel?Install CLI companion toolUnikraft InternalsRoadmap

Community

Events & MeetupsTalks & PresentationsPapers & PublicationsHackathonsContributingTeam

Features

PerformanceSecurityEfficiencyCompatibility

© 2024  The Unikraft Authors. All rights reserved. Documentation distributed under CC BY-NC 4.0.