Skip to content

docs: clustering release train — per-release R5 breakout #423

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
baron-3dl merged 1 commit into from
Aug 13, 2026
Merged

docs: clustering release train — per-release R5 breakout #423

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 20 additions & 0 deletions docs/release-roadmap-to-1.0.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,6 +87,26 @@ base executive):
| **DECnet Phase IV** | Clean-room DNA Phase IV; `SET HOST` + `NODE::` COPY both directions with a lab node | `vms-30e` | 1.0 |
| **Owns-its-kernel** | Kernel built from pinned source; VMS modules in-tree (`drivers/ovmx/`, `intree=Y` clears out-of-tree taint) + signed; curated config | `vms-19e` | 0.5 (prereq for DECnet + AXP) |

### R5 clustering train — per-release breakout (operator 2026-08-13)

R5 was too coarse ("0.4→1.0") to prioritize. The clustering work is broken into a
release train so each cut notches a concrete clustering win (per the release-dimensions
rule). Items carry an `rel-<milestone>` label so `rd list` shows what a cut owes.
"Full breadth + depth clustering, authentic" is make-or-break — this is its cadence.

| Rel | Clustering deliverable | Items (`rel-*` labelled) |
|---|---|---|
| **0.5** | **Cluster shows truth + reconfigures**: real membership everywhere; quorum-loss triggers reconfiguration; stop crashing peers; authenticated join | `vms-8d4` SHOW CLUSTER real membership · `vms-2d6` quorum-loss reconfiguration · `vms-953` peer CNXMGRERR bugcheck fix · `vms-405` group#/password auth · `vms-732` CLUSTER_AUTHORIZE nonce |
| **0.6** | **Distributed lock manager**: real cluster-wide locking behind RMS | `vms-7fa` DLM ENQ-class distributed lock state · `vms-407` RMS lock arbitrator (FAB/RAB) · `vms-551` cluster membership crosses into the executive |
| **0.7** | **Real-VAX interop hardening**: the RE friction the lab oracle exposes | `vms-257` REJECT-read-as-ACCEPT · `vms-298` stale send_seq/Con.ID reuse · `vms-abd` VAX refuses our DISCONNECT_REQ · `vms-770`/`vms-7f4` a61 audit gaps · `vms-41d` bind vote/quorum bytes |
| **0.8** | **Rejoin 5/5 (the white-whale) + diskless satellites** | `vms-4838`/`vms-2248` op 0x02 CM readmission on the member-initiated VC · `vms-2f3` rejoin 5/5 (ship at 4/5, never abandon 5/5 — operator 2026-08-13) · `vms-ce7` NISCS diskless satellite boot |
| **1.0** | **Join a real cluster + evacuate** — the takeover bar | `vms-ci` (epic: OVMX joins a real VMScluster) · `vms-600` real VAX MOUNTs OVMX-served MSCP · `vms-a662` genuine ODS-2 (cluster-interop gate) · live-workload evacuation, cluster stays up |

The `vms-2f3` thesis (operator 2026-08-13): building the forward features above — quorum
reconfiguration especially — is the most likely way to **surface the real root cause** of
the residual 1/5 rejoin loss, since reconfiguration and readmission share the
connection-manager machinery.

**Platform target set** (operator 2026-08-10, `vms-8ce`): **aarch64 + x86_64 + axp**;
VAX aspirational; ia64 excluded. AXP/Alpha is now first-class (supersedes CLAUDE.md
Rule 5's x86_64-primary framing — Rule 5 update owed). Design-cascade-sized; do not
Expand Down
Loading