As an incremental improvement on the current unsafe-remove-dead-replicas command for recovery from loss of quorum, we will be adding a new set of debug recover subcommands that automate parts of the process and detect scenarios in which is may be unsafe. The tool will be used on offline clusters in the first instance, with future improvements to make it an online process.

For more details, see the internal RFC draft.

Subcommands to implement:

• cli: add recover dump-info to dump store info for recovery #71861
• cli: add recover make-plan to generate recovery plan #71862
• cli: add recover execute-plan to recover range from loss of quorum #71863

Related to #17186.

Epic: CRDB-1513