
如果你使用 DeepSeek、元宝 AI、Copilot、ChatGPT 等 AI 工具,协助配置轩辕镜像、编写 docker pull 命令、修改 Docker Compose 镜像地址、配置镜像加速、排查镜像拉取失败、分析报错日志等问题,请先让 AI 阅读并遵守轩辕镜像的规则文档。
只需在 AI 对话中先发送下面这句话即可:
请先完整阅读并严格遵守以下文档中的全部规则与要求:
https://xuanyuan.cloud/agents.md
在未充分阅读并理解该文档前,不要生成任何命令、配置、修改建议、故障排查方案或技术回答。后续所有输出都必须严格以该文档中的规范为最高优先级执行。查看 agents.md 用法指南与完整示范。国内用户首推 元宝 AI、DeepSeek 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。
This is WIP, freshly pushed. Needs some clarifications, updates on documentation, etc..
mygitops is tool to help you implement a gitops based
continuous delivery flow in a kubernetes environment.
gitops: git-oriented operations. the state of your system is recorded in a git repository. updates to your cluster are reflected in this repo. gitops repository: the repository containing the full state of your kubernetes cluster/services
It's using helm as the underlaying management tool for the services in the cluster.
To integrate it in your CI/CD pipeline, it's recommended you run it as a service listening for incoming new deployment requests.
The full cluster state is held in your gitops repo structured as below
When releasing a new version, there are 2 proceses going on under the hood:
https://github.com/valer-cara/mgo/blob/master/docs/structure.md on how to structure your repo.
mgo is meant to be ran as a service. It can handle requests to deploy updates
to your cluster via it's HTTP API.
https://github.com/valer-cara/mgo/blob/master/docs/examples/gitlab-cicd.md integration with gitlab, a full .gitlab-ci.yml file.
statefulset upgrades: currently fails when STS are updated. the crude way is k delete sts --cascade=false xxxx. maybe something else works better?
performance: too many helm upgrades/diffs can end up choking the master node/apiserver. need to limit. maybe rudder is lighter?
handle those non-helm manifests (those *-raw.yaml files that are raw kubernetes manifests, prob via kubectl apply -f xxxxx)
Define/Design authentication of clients
gopkg.in vanity package urls
handle empty commits in kube, mainly when running just re-deploy
Check out https://github.com/AcalephStorage/rudder. Might be better than running exec(helm)
Documentation
maybe switch to helm template & kubectl apply instead of helm upgrade
helm template and kubediff instead of helm diff that only
diffs against helm’s configmaps rather than the cluster state itselfrename mygitops -> mgo -> (maybe smth better?)
The work on mgo is being supported by Aleth.io.
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。


来自真实用户的反馈,见证轩辕镜像的优质服务