This is a tracking issue for decommissioning improvements in 22.2.

Benchmarking

• Create benchmarks for decommissions in relevant cases
• Evaluate benchmarks against optimal time estimates

Observability

• Export queue lengths and snapshots in progress via metrics
• Add tracing for errors during decommission
• Track error counts for a decommission
• Snapshot Dashboard showing: Snapshots in progress, queue lengths, purgatory counts, aggregate error counts.
• Avoid displaying retired nodes in Admin UI

Operational Improvements

• Receiver-side snapshot queueing to ensure fairness (kvserver: fairly prioritize snapshot application #86799)
• New allocator "destination store selector" for decommissioning to pick from available nodes rather than best node (Recover replicas to a "good enough" store instead of the "best" store #86265)
• Fix requeue in replicate queue to downprioritize range count rebalance actions during a decommission

Jira issue: CRDB-18247

Epic CRDB-14621