Skip to main content

Docker Images Onboard

The following table contains the list of available Docker images onboard. To locally pull and test these images, run the following:

docker pull <Image Tag>

For example, for alpine, run the following (select arm64 version also):

docker --platform=arm64 pull alpine:3

Just be aware that those images were pulled a while ago, the latest tags for example are not from now, but from the time pulled. To be sure your workload will run on the FM, please assure you had a testing in our Engineering Model first. The latest, alpine, slim tags were push as-is, you will find in () the version running. In the registry, the tag does not contain the () part. For example, if you want to use the image alpine on the latest(3.20) you will use FROM fpga.dphi.space:5000/alpine:latest. For your reference and internal testing, you can fix it to version 3.20 in case the latest tag has been otherwise moved.

Image NameTag
almalinux8
almalinux9
alpine3.18
alpine3.19
alpineedge(20250108)
alpinelatest(3.22.1)
bellsoft/liberica-runtime-containerjre-21-slim-musl
busyboxlatest(1.37.0)
busyboxmusl(1.37.0-musl)
busyboxstable(1.36.1)
caddy2
caddyalpine(2.10.2-alpine)
cassandra3.11
cassandra4
debianbookworm
debianbookworm-slim
debianbullseye
debianbullseye-slim
dinov3-sat493m_orin16-r36.4.0(*)latest
distroless-baselatest(base)
distroless-java17-debian12latest(java17-debian12)
distroless-nodejs20-debian12latest(nodejs20-debian12)
distroless-python3-debian12latest(python3-debian12)
distroless-staticlatest(static)
dockercli
dockerdind
dockerlatest(28.4.0)
dotnet-aspnet8.0
dotnet-runtime8.0
dotnet-sdk7.0
dotnet-sdk8.0
dphi-embedded-ml-armv8-py311(**)latest
eclipse-temurin11-jdk
eclipse-temurin17-jdk
eclipse-temurin17-jre
eclipse-temurin21-jdk
eclipse-temurin21-jre
elasticsearch7.17.0
elasticsearch8.11.0
fedora38
fedora39
fedoralatest(42)
fluent/fluentdlatest(v1.19.0-1.0)
gcc12
gcc13
gcclatest(15.2.0)
golang1.20
golang1.20-alpine
golang1.21
golang1.21-alpine
golanglatest(1.25.1)
graalvm-native-image-community21
gradle8.5-jdk17
gradle8.5-jdk21
grafana/grafana9.3.6
grafana/grafanalatest(12.2.0-16791878397)
haproxy2.8
haproxyalpine
httpdalpine
httpdlatest(2.4.65)
influxdb1.8
influxdb2.7
l4t-mllatest(r36.4.4)
l4t-pytorch-r36-2-0latest(r36.2.0)
l4t-pytorch-r36-4-0latest(r36.4.0)
mariadb10.11
mariadb10.6
mariadb11
mariadb11.8.2
mariadblatest(11.8.2)
mcr.microsoft.com/dotnet/aspnet8.0
mcr.microsoft.com/dotnet/runtime8.0
mcr.microsoft.com/dotnet/sdk7.0
mcr.microsoft.com/dotnet/sdk8.0
mysql8.0
mysqllatest(9.4.0)
nginxalpine(1.29.0-alpine)
nginxlatest(1.29.0)
node18
node18-alpine
node18-slim
node20
node20-alpine
node20-slim
nodelts(22.19.0)
nodelts-alpine(22.19.0-alpine3.22)
php8.1-fpm
php8.2-apache
php8.2-fpm
php8.3-apache
php8.3-cli
php8.3-fpm
postgres13
postgres14
postgres15
postgres15-alpine
postgres16
postgres16-alpine
postgres17
postgreslatest(17.6-trixie)
prom/node-exporterlatest(v1.9.1)
prom/prometheuslatest(v3.5.0)
prom/prometheusv2.42.0
python3.10
python3.10-slim
python3.11
python3.11-alpine
python3.11-slim
python3.12
python3.12-alpine
python3.12-slim
redis6-alpine
redis7-alpine
redislatest(8.2.1-bookworm)
registry2
rocker/rstudio4.3.2
rockylinux8
rockylinux9
ruby3.2
ruby3.2-alpine
ruby3.3
ruby3.3-alpine
rubylatest(3.4.5-trixie)
rust1.75
rust1.75-slim
rustalpine(1.89.0-alpine)
rustlatest(1.89.0)
traefiklatest(v3.5.2)
traefikv3.0
traefikv3.5.2
ubuntu20.04
ubuntu22.04
ubuntu24.04
ubuntulatest(24.04)
ultralytics-latest-jetson-jetpack6latest(8.3.241-jetson-jetpack6)

(*) dinov3 is a model developed by Meta (https://ai.meta.com/dinov3/). It encodes the image into embedding s just like LLMs do so for test, which can later be used for many tasks: segmentation, depth estimation, classification, change detection etc.. We loaded the model dinov3-vitl16-pretrain-sat493m which a backbone thats pretrained on satellite imagery. It is therefore ideal for usage on CG2 and later missions with high resolution images. The dockerfile requires a hugging face token to pull the dinov3-vitl16-pretrain-sat493m which is removed from this Dockerfile

(**) dphi-embedded-ml is an image that has useful machine learning libraries that can run on a cpu (torch, tflite, onnx runtime). It is a lightweight alternative for images like dustynv-ml that can only run on the jetson. The image also includes useful libraries for orbital calculations (sgp4, pyorbital, pyproj). This is an ideal compromse to test AI in a CPU rather than on the GPU. You can pull it from dockerhub or you can find the Dockerfile used at: Dockerfile.