Docs › Daily use
Daily use for DevOps engineers
Pipelines, runners, environments, variables, merge gates and releases.
Pipelines
- A pipeline lives in
.mcdock/pipeline.ymlin the repository, so it is reviewed like code. - Use stages and jobs, branch filters, manual jobs, timeouts and job variables.
- Every run shows live logs, keeps artifacts, and reads JUnit test reports, so failed tests show on the pull request.
stages: [build, test, deploy]
jobs:
build:
stage: build
runs-on: [windows] # runner labels
script:
- dotnet build -c Release
artifacts: [bin/Release/**]
tests:
stage: test
script:
- dotnet test --logger "junit;LogFilePath=results.xml"
junit: [**/results.xml]
deploy-production:
stage: deploy
environment: production
branches: [main]
when: manual
script:
- ./deploy.ps1 -Server $env:PROD_SERVER
Job settings: stage, runs-on, shell, script, branches, tags: true, when (on_success, manual, always), environment, junit, artifacts, timeout (minutes, default 60), variables and allow-failure.
Runners
- A runner is the same
mcdock.exe, run as a runner on any Windows or Linux machine. It connects outbound to the server, so you don't open any firewall ports to it. - Jobs run in PowerShell on Windows, and also in
cmd,pwsh,bashorsh. Give each runner only the tools and rights its jobs need. - Check Admin → Runners each day. When a runner is offline, its jobs wait. To add a runner, give it a name and labels such as
windows, dotnet, then runmcdock.exe runner --url <server> --token <token>on the build machine.
Environments and approvals
- Create test, staging and production environments. For each one, set the allowed branches and its approvers.
- Turn on the two-person rule for production, so the person who starts a deployment can't approve it.
- Keep secrets in repository → Environments → CI/CD variables. Masked values show as
[MASKED]in logs, protected variables only reach protected branches, and an environment scope limits a secret to that deployment.
Merge gates
- On the protected branch, add
mcdock/pipelineas a required status check, require the CODEOWNERS approvals, and dismiss stale approvals. - Merge when ready merges by itself once everything passes. This keeps the main branch green without anyone watching.
Every day
- Look at the failed runs and fix flaky tests. Don't just press Retry.
- Watch the build times. Cache dependencies on the runners and split slow jobs.
- Use webhooks and the REST API, with scoped and expiring tokens, to connect your other tools.
Releases
- Tag a version, let the pipeline build it, and publish it under Releases with notes and artifacts.
Updated 21 Nov 2026 · Something missing? Ask in the community.