Skip to content
Browse documentation

Infrastructure

Deployment patterns for different projects

Choose reviewed operations for containers, application services, clusters, static sites and database migrations.

Choose operations from your project’s release process. These examples assume existing infrastructure and reviewed tooling, not a particular operating-system distribution, hosting provider or web server. Vopit does not provision a universal deployment environment. Every path, identity and health check must come from the project you opened.

Project / targetHow to connectEvidence to keep
Docker Compose containersSSH to a POSIX host with the project’s container tools already installed.Image digests, validated configuration, health checks and previous image references.
Python, Go or Node application serviceSSH to the project’s approved release scripts.Artifact checksum, runtime compatibility, service status and public response.
Kubernetes applicationSSH to an authorized runner with scoped cluster tooling.Namespace, manifest revision, rollout status and public service result.
Static site or object storageSSH to an approved publishing runner or a wrapper for the existing CI pipeline.Output manifest, cache behavior and public asset checks.
Windows/IIS applicationUse an existing deployment pipeline exposed through an approved POSIX runner.Pipeline result, deployed version and application checks; no native Windows SSH adapter is available.
PostgreSQL or MySQL/MariaDB databaseDatabase connector for reviewed SQL; SSH for an installed project migration tool.Backup identifier, migration version and compatibility with application releases.

Configure the target and operations in Vopit

  1. In Settings → Connections → Infrastructure, add the actual SSH or database target. Verify its identity and import credentials through the dedicated controls.
  2. For SSH, set the working root and public URL appropriate to the project. Create the upload parent directory through your normal server administration procedure.
  3. Add separate operations for the required release steps. Set an ID, label, purpose, execution type, policy and timeout for each. Review paths and commands before saving.
  4. Assign the saved target to the DevOps agent. If generating operations with AI, enable relevant project excerpts and review the listed sources and assumptions.
  5. Start with a bounded inspection. Then deploy an identified artifact or image, verify the result, and keep evidence of the previous working release.

Application services: project-owned release scripts

For Python, the project may package a wheel and manage a virtual environment. For Go, it may deploy a compiled binary. For Node, it may install locked runtime dependencies. Configure the runtime and service controls in the project’s own reviewed scripts; an agent’s name does not select a build system.

Illustrative names for three separate SSH operations — these scripts must already exist and be reviewed
./automation/activate-release.sh
./automation/check-health.sh
./automation/restore-release.sh

These are operation examples, not scripts supplied by Vopit. The remote account needs only the privileges each operation requires. A rollback script must identify the previous release and state what happens to persistent data.

Containers: identify the image before activation

  1. Record the reviewed image digest and container configuration. Keep registry credentials in the approved runner’s credential store.
  2. Validate the project’s Docker Compose configuration through a saved operation, then activate the intended image references.
  3. Check application health and the externally visible version. Keep persistent volumes when returning to a previous image; data restoration is a separate operation.

Clusters: use a scoped runner

For Kubernetes, give the runner only the namespace and actions required by the project. Keep cluster credentials out of prompts and uploads. Save operations for manifest validation, application, rollout checks and recovery. Verify the public service in addition to readiness. Vopit has no native Kubernetes API connector; the installed runner tools perform those actions.

Static hosting and existing CI pipelines

Build with the project’s actual toolchain and publish only its output. An approved runner can invoke installed hosting tools or trigger a reviewed CI workflow. Configure the destination and cache policy in that workflow. Vopit has no built-in S3 connector, hosting-provider deployment or native CI deploy; publishing needs an explicitly configured runner operation. Cranks can complement it: the Vercel app lists and inspects deployments (read-only), the GitHub and GitLab apps work with issues, an HTTP / Custom API action can call a CI webhook, and an Infrastructure operation Crank runs a saved operation, for example in a trigger sequence. Confirm the deployed revision and public assets after the pipeline completes.

Routing and certificates

Add distinct operations for configuration validation, applying routing changes, certificate renewal and HTTPS verification when your project needs them. Use the tools already responsible for that environment. Review the affected domains and services; a successful renewal command alone does not prove that clients receive the new certificate.

Database migrations and restore plans

  1. Choose the database connector for exact reviewed SQL, or an SSH operation for the project’s installed migration tool. Use a restricted database role and verified TLS.
  2. Record a tested backup or recovery point and check compatibility between the old and new application versions before migration.
  3. Review the exact migration, set its timeout, and approve execution. Check the resulting schema/version and the application afterwards.
  4. If the release fails, follow the project’s recovery plan. A code rollback cannot guarantee reversal of schema or data changes.

All documentation