如果你用 DeepSeek、元宝AI、Copilot 等 AI 代为编写轩辕镜像的 docker pull、修改 compose 镜像地址,或排查拉取失败等问题,请让 AI 先读规则再动手,只要在 AI 对话里直接发送下面这句话即可:
生成 docker pull、修改 compose 的 image,或排查拉取失败前,请先阅读并严格遵守 https://xuanyuan.cloud/agents.md 中的规则。国内用户首推 DeepSeek、元宝 AI 的深度思考模式,不推荐豆包 AI;Cursor 等编辑器可在对话 @ 该链接,或加入 User Rules。 若 AI 无法访问外链,可 打开说明文档 复制全文粘贴。文档会随站点更新,复制内容可能过期,建议定期检查。
Praqma/Network-Multitool is now wbitt/Network-MultitoolFew years ago, I created this tool with Henrik Høegh, as praqma/network-multitool. Praqma was bought by another company, and now the "Praqma" brand is being dismantled. This means the network-multitool's git and docker repositories must go. Since, I was the one maintaining the docker image for all these years, it was decided by the current representatives of the company to hand it over to me so I can continue maintaining it. So, apart from a small change in the repository name/url, nothing has changed.
The existing/old/previous container image praqma/network-multitool will continue to work and will remain available for "some time" - may be for a couple of months - not sure though.
Or:
docker pull wbitt/network-multitool
A (multi-arch) multitool for container/network testing and troubleshooting. The main docker image is based on Alpine Linux. There is a Fedora variant to be used in environments which require the image to be based only on RedHat Linux, or any of it's derivatives.
The container image contains lots of tools, as well as a nginx web server, which listens on port 80 and 443 by default. The web server helps to run this container-image in a straight-forward way, so you can simply exec into the container and use various tools.
Openshift is very strict about how a container image should run. So, the openshift variant of the multitool has the following limitations / changes:
traceroute, tcptraceroute, etc, will not work)1180 and 11443 - not 80 and 443setuid, so those tools remain usable. Tools set with setuid are:
Remember, this multitool is purely a troubleshooting tool, and should be used as such. It is not designed to abuse openshift (or any system's) security, nor should it be used to do so.
80, port 443) - with customizable ports!Size: 16 MB compressed, 38 MB uncompressed
All tools from "minimal", plus:
Size: 64 MB compressed, 220 MB uncompressed
/bin/sh shell interpreter - not /bin/bashSize: 72 MB uncompressed
Note: The SSL certificates are generated for "localhost", are self signed, and placed in /certs/ directory. During your testing, ignore the certificate warning/error. While using curl, you can use -k to ignore SSL certificate warnings/errors.
$ docker run -d praqma/network-multitool
Then:
$ docker exec -it container-name /bin/bash
Create single pod - without a deployment:
$ kubectl run multitool --image=praqma/network-multitool
Create a deployment:
$ kubectl create deployment multitool --image=praqma/network-multitool
Then:
$ kubectl exec -it pod-name /bin/bash
Note: You can pass additional parameter --namespace=<your-desired-namespace> to the above kubectl commands.
$ oc new-project test-project-1 $ oc new-app praqma/network-multitool:openshift --name multitool-openshift $ oc status $ oc get pods $ oc logs pod-name $ oc exec -it pod-name /bin/sh $ oc port-forward pod-name 1180:1180 11443:11443
Sometimes you want to do testing using the host network. This can be achieved by running the multitool using host networking.
$ docker run --network host -d praqma/network-multitool
Note: If port 80 and/or 443 are already busy on the host, then use pass the extra arguments to multitool, so it can listen on a different port, as shown below:
$ docker run --network host -e HTTP_PORT=1180 -e HTTPS_PORT=11443 -d praqma/network-multitool
For Kubernetes, there is YAML/manifest file multitool-daemonset.yaml in the kubernetes directory, that will run an instance of the multitool on all hosts in the cluster using host networking.
$ kubectl apply -f kubernetes/multitool-daemonset.yaml
Notes:
--namespace=<your-desired-namespace> to the above kubectl command.daemonset is configured to run multitool on port 1180 and 11443. You can change this in the YAML file if you want.There are times when one may want to join this (multitool) container to another container's IP namespace for troubleshooting, or on the host network. This is true for both Docker and Kubernetes platforms. During that time if the container in question is a web server (nginx, apache, etc), or a reverse-proxy (traefik, nginx, haproxy, etc), then network-multitool cannot join it in the same IP namespace on Docker, and similarly it cannot join the same pod on Kubernetes. This happens because network multitool also runs a web server on port 80 (and 443), and this results in port conflict on the same IP address. To help in this sort of troubleshooting, there are two environment variables HTTP_PORT and HTTPS_PORT , which you can use to provide the values of your choice instead of 80 and 443. When the container starts, it uses the values provided by you/user to listen for incoming connections. Below is an example:
$ docker run -e HTTP_PORT=1180 -e HTTPS_PORT=11443 \ -p 1180:1180 -p 11443:11443 -d local/network-multitool 4636efd4660c2436b3089ab1a979e5ce3ae23055f9ca5dc9ffbab508f28dfa2a $ docker ps CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES 4636efd4660c local/network-multitool "/docker-entrypoint.…" 4 seconds ago Up 3 seconds 80/tcp, 0.0.0.0:1180->1180/tcp, 443/tcp, 0.0.0.0:11443->11443/tcp recursing_nobel 6e8b6ed8bfa6 nginx "nginx -g 'daemon of…" 56 minutes ago Up 56 minutes 80/tcp nginx $ curl http://localhost:1180 Praqma Network MultiTool (with NGINX) - 4636efd4660c - 172.17.0.3/16 - HTTP: 1180 , HTTPS: 11443 $ curl -k https://localhost:11443 Praqma Network MultiTool (with NGINX) - 4636efd4660c - 172.17.0.3/16 - HTTP: 1180 , HTTPS: 11443
If these environment variables are absent/not-provided, the container will listen on normal/default ports 80 and 443.
Well, normally, if a container does not run a daemon/service, then running it (the container) involves using creative ways / hacks to keep it alive. If you don't want to suddenly start browsing the internet for "those creative ways", then it is best to run a small web server in the container - as the default process.
This helps you when you are using Docker. You simply execute:
$ docker run -d praqma/network-multitool
This also helps when you are using kubernetes. You simply execute:
$ kubectl run multitool --image=praqma/network-multitool
The multitool container starts as web server - so it remains UP. Then, you simply connect to it using:
$ docker exec -it some-silly-container-name /bin/sh
Or, on Kubernetes:
$ kubectl exec -it multitool-3822887632-pwlr1 -- /bin/sh
This is why it is good to have a web-server in this tool. Hope this answers the question! Besides, I believe that having a web server in a multitool is like having yet another tool! Personally, I think this is cool! Henrik thinks the same!
We have tried to put in all the most commonly used tools, while keeping it small and practical. We can't have all the tools under the sun, otherwise it will end up as something like this.
However, if you have a special need, for a special tool, for your special use-case, then I would recommend to simply build your own docker image using this one as base image, and expanding it with the tools you need.
There is absolutely no need to use LetsEncrypt. This is a testing tool, and validity of SSL certificates does not matter.
One could argue that it is possible to simply install the tools on the hosts and get over with it. However, we should keep the infrastructure immutable and not install anything on the hosts. Ideally we should never ssh to our cluster worker nodes. Some of the reasons are:
daemonset, it makes it easier to integrate with other resources. e.g. Use volumes for packet capture files, etc.daemonset provides a 'cloud native' approach to provision debugging/testing tools.exec into the daemonset, without needing to SSH into the node.Contributions are welcome for packages/tools ***ed "absolutely necessary", of "core" nature, are "minimal" in size, and "have large number of use-cases". Remember, the goal is not to create yet another Linux distribution! :)
您可以使用以下命令拉取该镜像。请将 <标签> 替换为具体的标签版本。如需查看所有可用标签版本,请访问 标签列表页面。



探索更多轩辕镜像的使用方法,找到最适合您系统的配置方式
通过 Docker 登录认证访问私有仓库
发给 Cursor、ChatGPT、豆包等 AI 的说明文档
无需登录使用专属域名
Kubernetes 集群配置 Containerd
K3s 轻量级 Kubernetes 镜像加速
VS Code Dev Containers 配置
Podman 容器引擎配置
HPC 科学计算容器配置
ghcr、Quay、nvcr 等镜像仓库
Harbor Proxy Repository 对接专属域名
Portainer Registries 加速拉取
Nexus3 Docker Proxy 内网缓存
需要其他帮助?请查看我们的 常见问题Docker 镜像访问常见问题解答 或 提交工单
docker search 限制
站内搜不到镜像
离线 save/load
插件要用 plugin install
WSL 拉取慢
安全与 digest
新手拉取配置
镜像合规机制
不支持 push
manifest unknown
no matching manifest(架构)
invalid tar header(解压)
TLS 证书失败
DNS 超时
域名连通性排查
410 Gone 排查
402 与流量用尽
401 认证失败
429 限流
D-Bus 凭证提示
413 与超大单层
来自真实用户的反馈,见证轩辕镜像的优质服务