Skip to content

Configuration

Simplyblock Disaster Recovery is configured entirely through Kubernetes custom resources on the hub cluster, in the API groups dr.simplyblock.io/v1alpha1 and sitemap.simplyblock.io/v1alpha1, by kubectl or through the Control Center. The resources describe which sites exist, how data moves between them, in which directions applications may move, and how each application is started on the target site.

Configuration Model

The configuration is layered. Each layer references the one above it:

  1. DRConfig: The cluster-scoped singleton default holds hub-wide settings: the S3 archive and state bundle, the retention of finished runs, the agent settings (including the image allow list for hook Jobs), and feature gates. It is created by the hub chart (see Installing the Hub).
  2. ProtectionPlan: A cluster-scoped plan declares a set of sites, the protected StorageClasses, the replication methods (sync, async, or snapshot-s3), and the S3 stores of the sites. See Protection Plans and Replication Types.
  3. DRPath: A cluster-scoped path declares one direction between two sites of a plan and the actions allowed along it (Failover, Relocate, Test). Directions are declared, never inferred. See DR Paths.
  4. ProtectedApplication: A namespaced resource binds one application to a plan, with a source site, one target site, and a method. It also declares health probes and dependencies. See Protected Applications.

  5. Tiers, hooks, and Recipes: Within a protected application, tiers define the boot order on the target site, and external hooks run steps outside the cluster, such as DNS changes. A hand-written Recipe can replace the tiers when more control is needed. See Workflows and Recipes.

  6. Site profiles and mappings: In the API group sitemap.simplyblock.io/v1alpha1, a SiteProfile per managed cluster describes the site and binds logical roles to its networks, and a DHCPServer names a DHCP server dr-hub renders reservations into. The site mapper translates the network attachments and guest addresses of virtual machines between the sites of a path. See Site Profiles and Mappings.

Operational resources, such as RecoveryAction, RecoveryPlan, TestBubble, TestSchedule, and RestoreAction, build on this configuration and are described in Operations and Testing.

Resource Overview

Kind Scope Short name Written by
DRConfig Cluster (named default) None dr-admin
ProtectionPlan Cluster pplan dr-admin
DRPath Cluster None dr-admin
ProtectedApplication Namespaced papp dr-operator
SiteProfile Cluster (one per cluster) None dr-admin
DHCPServer Cluster None dr-admin

The relationships between the resources are:

Relationships between the DR resources
DRConfig "default"
ProtectionPlan "fra" ── sites: fra-a, fra-b ── methods: async-5m
 ├── DRPath "fra-a-to-fra-b"  (Failover, Relocate, Test)
 ├── DRPath "fra-b-to-fra-a"  (Relocate)
 └── ProtectedApplication "orders"  (source fra-a, target fra-b, method async-5m)
      ├── tiers
      └── health.probes, externalHooks

The concepts behind the resources are explained in Protection Plans and Paths and Protected Applications.

Pages in This Section