MY TOOLBOX.NET / C#AzureDockerGitHub ActionsTerraformCloudflareArgoCDLinux
⎇
GitHub
Source control & Actions CI/CD for every push
▦
Terraform
Infrastructure as code for Azure and the homelab
⟳
ArgoCD
GitOps delivery for Kubernetes workloads
☁
Cloudflare
Tunnels and DNS in front of everything I host
◱
Docker
Packaging for the site, the sandbox and the lab
△
Azure
AKS, PostgreSQL and identity for the sandbox
01 / Sandbox
A place to run things.
Infrastructure, experiments and useful side projects.
Infrastructure / Azure / v3
Azure platform sandbox
Keep the data. Rebuild the cluster. Pay for usage, not an always-on platform.
A Terraform platform for running small Kubernetes workloads on Azure with a clear separation between persistent services and disposable compute. The sandbox is available at alau.work.
Provisioning order: foundation → data → compute. Terraform manages Azure; application charts and ingress are separate.
Private by default
PostgreSQL is private inside the VNet. Applications use workload identity and DefaultAzureCredential, not storage keys.
Scoped access
GitHub Actions uses OIDC for the protected main branch. Entra RBAC handles operator and application access.
02 / Lifecycle
Build it. Use it. Turn it off.
01 / Push to mainValidate + apply
Apply data, then compute.
02 / Start sessionStart + deploy
Connect with Entra and run workloads.
03 / Stop sessionKeep the data
Stop worker compute and retain services.
04 / Optional cleanupDelete compute
Remove AKS and keep the data layer.
A later push recreates compute. Full data deletion is a separate confirmed action.
Automation with guardrails
Creation, deletion and session workflows share one concurrency group. A daily stop runs at 20:17 UTC with a retry at 21:17 UTC.
Cost-aware by design
Stopped AKS still retains disks and networking. Compute deletion removes the cluster while PostgreSQL, blobs and secrets remain available.
03 / At a glance
The important numbers.
Estimated monthly cost
$85–95 before tax, based on East US retail estimates and up to 60 AKS running hours. An always-on cluster would be roughly $310/month before variable usage.
I’m Alau Bolatov, a DevOps engineer and software developer working with .NET, Azure, and infrastructure automation. I like understanding how things fit together, from the code to the container to the machine it runs on.