List view

• Upgrade Phase 3: zero-downtime HA (replica failover)

  The RPO=0 end-state for clustered deployments: upgrade the in-sync raft replica, promote (committed-log fence), upgrade the old primary. Needs prod running clustered.

  No due date

  •0/1 issues closed
  0% complete1 open0 closed

• Upgrade Phase 2: seamless (zero-downtime single-node)

  Remove the upgrade downtime on a single node: systemd socket-activation (no connection-refused) + a fast state handoff (tmpfs / streamed old->new) with a -LOADING cutover freeze.

  No due date

  •0/3 issues closed
  0% complete3 open0 closed

• Upgrade Phase 0-1: verified self-update + data-safe swap

  ironcache upgrade MVP: the verified self-updater spine (fetch/verify/atomic-swap/health-gate/auto-rollback) + a data-safe snapshot-reload swap with a write-freeze. Works on the single-node prod box; loss-free, brief downtime.

  No due date

  •2/4 issues closed
  50% complete2 open2 closed

• Implementation Readiness

  Critical-path gate: these issues must be resolved or fully specified before the first implementation PR merges. Sequence is in docs/ROADMAP.md and the roadmap tracking issue; waves are the wave:0..3 labels.

  No due date

  •43/43 issues closed
  100% complete0 open43 closed

• M2: Prototype-Ready Design

  Decisions resolved to the point a prototype can be built.

  No due date

  •58/58 issues closed
  100% complete0 open58 closed

• M1: Architecture Specification

  The subsystem design specifications and their decisions.

  No due date

  •47/47 issues closed
  100% complete0 open47 closed

• M0: Vision and Scope

  Tenets, scope, committed non-goals, and the research foundations.

  No due date

  •17/17 issues closed
  100% complete0 open17 closed