ADR 0022 — v6 fleet membership and future-DR naming

Active fleet is hub-dc-v6 + spoke-dc-v6 only; pre-v6 DR clusters decommissioned; hub-dr-v6 / spoke-dr-v6 are reserved names. Supersedes the cluster-list portion of ADR 0001.

Date: 2026-05-10 Status: Accepted. Supersedes the cluster-list portion of ADR 0001.

Context

ADR 0001 (2026-05-04, “Maintain A Local Operator Workspace”) was written when the lab fleet was the pre-v6 cluster set:

• hub-dc — active management hub.
• spoke-dc — active workload cluster.
• hub-dr — management standby.
• spoke-dr — workload standby.

Since 0001 was accepted, the fleet has been substantially rebuilt:

• The pre-v6 hub-dr and spoke-dr clusters are decommissioned. They are not in active or standby scope. Their cluster inventory was removed from CLUSTERS.md together with this ADR.
• The pre-v6 hub-dc and spoke-dc clusters are transitional — still inventoried in CLUSTERS.md for context but not the active GitOps target.
• The v6 rebuild clusters hub-dc-v6 and spoke-dc-v6 are now the active fleet. They are managed via the federated GitOps repository (comptech-platform/openshift-ops/openshift-platform-gitops on internal GitLab) per ADR 0015 + ADR 0018.
• Future DR clusters under v6, if/when introduced, will follow the v6 naming as hub-dr-v6 and spoke-dr-v6. They are not currently in scope.

The risk this ADR fixes is silent drift between docs: agents and operators continuing to write hub-dc / spoke-dc in new plans, scripts, runbooks, and session reports because the original ADR named them as the active fleet. Without an explicit supersession, every new doc starts from the wrong baseline.

Decision

The active OpenShift fleet membership for the opp-full-plat workspace is:

This ADR supersedes the cluster-list portion of ADR 0001 (the bullet list naming hub-dc / spoke-dc / hub-dr / spoke-dr as the four-cluster fleet). ADR 0001’s other decisions — maintain a local operator workspace, keep AGENTS.md / CLUSTERS.md / etc. — remain in force. Only the cluster-membership statement is updated.

When operating against the fleet:

• “Active management” or “the hub” means hub-dc-v6.
• “Active workload” or “the spoke” means spoke-dc-v6.
• The pre-v6 hub-dr and spoke-dr cluster names must not appear in new manifests, plans, scripts, runbooks, or session reports.
• The pre-v6 hub-dc and spoke-dc cluster names are acceptable only when referring to historical / transitional context; new active work targets the v6 rebuild clusters.
• hub-dr-v6 / spoke-dr-v6 may appear in forward-looking planning text (e.g., “future cluster paths”) but no live work targets them until DR is explicitly reintroduced through a tracked issue and ADR.

Alternatives considered

Amend ADR 0001 in place. Edit the original bullet list to reflect the v6-only fleet. Rejected because ADRs are immutable historical records; amending them in place loses the trail of “what was true on 2026-05-04 vs. 2026-05-10.” The convention is to write a new ADR that explicitly supersedes the old one and to leave the old text intact.

Write the supersession as a paragraph in CURRENT_STATE.md only. Lighter ceremony. Rejected because:

• CURRENT_STATE.md rotates; it’s a living snapshot, not a durable decision.
• A future operator reads ADRs at session start. Without an ADR, the supersession lives only in chat logs and personal notes.
• The drift problem the ADR is fixing is exactly that operators write new things using the old fleet names; only an ADR-level supersession is durable enough to prevent that.

Leave the DR pair documented as “decommissioned standbys” but reserve them in case we rebuild. Rejected because:

• The DR pair is not just “decommissioned” — the underlying VMs and physical workers are gone or repurposed.
• Reserving the exact pre-v6 names would re-introduce ambiguity with the v6 active clusters.
• The future-DR naming follows v6 convention (hub-dr-v6 / spoke-dr-v6) so there’s no confusion.

Use semantic versioning suffixes (hub-dc-v7, etc.) instead of -v6. Rejected because the lab’s existing rebuild is named “v6,” and renaming the suffix would require updating GitOps paths, ACM ManagedCluster names, certs, DNS, and operator memory. Continuity wins.

Validation

Closeout actions taken at acceptance:

• opp-full-plat/CLUSTERS.md — hub-dr and spoke-dr sections removed; Active Scope updated to name hub-dc-v6 / spoke-dc-v6.
• opp-full-plat/plans/federated-gitops-readiness-gates.md — references to “future hub-dr” / “future spoke-dr” updated to “future hub-dr-v6” / “future spoke-dr-v6”.
• ADR Index issue (#131) — this ADR added with a note on ADR 0001 that the cluster list is superseded by 0022.
• ADR 0001 itself is NOT modified. Historical record preserved.

Consequences

• Future readers see ADR 0022 as the source of truth for active fleet membership; ADR 0001 is the historical anchor for the workspace decision.
• Documents and code that still mention old hub-dr / spoke-dr are flagged as drift-from-current-state and should be updated when touched. A drift-grep over the workspace is part of routine session closeouts.
• Reintroducing DR under v6 (hub-dr-v6 / spoke-dr-v6) requires a new ADR or amendment to record the install, ACM ManagedCluster registration, and platform-gitops clusters/<name>/ directory structure. This ADR reserves the names; it does not authorize the build.

References

• Source: opp-full-plat/adr/0022-v6-fleet-membership.md
• Superseded predecessor (cluster-list portion only): ADR 0001
• Federation context: ADR 0015
• v6 management baseline: ADR 0018
• Platform admin handoff: opp-full-plat/connection-details/platform-admin-handoff.md
• Compliance implementor handbook: opp-full-plat/connection-details/compliance-implementor-handbook.md
• Tracker: zeshaq/opp-full-plat #156 (OPS-V6-FLEET-1 — the purge work)
• ADR Index issue: zeshaq/opp-full-plat #131