Availability

Start with Availability modes to configure none, dc, or ha. The remaining pages define durability contracts and operating procedures.

• Availability contracts

  The durability and failure guarantees of none, dc, and ha modes.

• Availability deployment and sizing

  Size and deploy none, dc, and ha modes.

• Availability modes

  Configure and operate none, dc, and ha modes.

• Node liveness

  Track datacenter replica health from bounded heartbeats while leaving the configured roster unchanged.

• Authority transfer and drain

  Transfer an authority after a confirmed home failure and drain retained writes.

• Derived-view frontiers

  Gate replica reads until each required view reflects the applied metadata serial.

• Finalizing admitted content

  Publish admitted content at its authority home with fencing, validation, and idempotent outcomes.

• Home assignment on first publish

  Assign one authority home through the control quorum and fence later writes with its epoch.

• Availability control listener

  Configure the private availability control socket.

• Filesystem placement reconciliation

  Verify blob placements, retire out-of-policy copies, and schedule repairs.

• Planned authority transfer

  Move a healthy authority after its target catches up.

• Blob reclamation

  Fence deletion of unreferenced content behind placements and replication frontiers.

• Managing voting membership

  Add, promote, replace, or remove voters in the ownership consensus group.

• Observe availability health

  Inspect topology, placement, operations, readiness, and replica frontiers.

• Rolling upgrade and rollback

  Preflight, order, and recover a rolling availability upgrade.

• Upgrade and roll back an availability cluster

  Upgrade or roll back a live dc or ha cluster one build at a time.

• Version compatibility and rolling upgrades

  Negotiate mixed-version clusters and enforce rollback limits.

• Failover and recovery

  Classify availability failures, recover the selected mode, and verify service before restoring traffic.

• Client behavior across availability modes

  Retry, fencing, and read consistency in none, dc, and ha modes.

• Visibility replication

  Replicate ordered visibility transitions and retain their tombstones.