seg backup/mirror command

[sudeepdino008]

source: https://discord.com/channels/687972960811745322/886550095749873666/1452020965570515107

essentially, create some kind of backup of datadir. Most of the data is in immutable snapshots, so we should be able to use hardlinks for cheap backups.

• devs/devops sometimes creates backups of files before attempting some commands, specially on new servers.
• useful for dev workflows - create backup, execute some offline commands on it to get it to a desired datadir state from which some bug can be reproduced, then create multiple backups of this backup to be able to test things on a particular state.
• users are not always aware of the different offline commands/recovery strategies, and often shoot themselves on foot - they do something wrong or it's not easy to test commands out ("my datadir works, why mess with it?"). Giving them a way to do backups can provide a safer env to try out things, debug things more independently etc.

[wmitsuda]

FYI, re: your mirror-datadir script, if you are interested, on mac/APFS, I simply use: cp -cR datadir-orig datadir-new.

-c does copy on write for each file individually, but that doesn't matter much bc snapshots are immutable, so at FS level they are pointed to the same thing always. and other files are cloned at first write automatically, so you inherit p2p DB, etc. as well and don't risk changing the original datadir.

what I do is to keep some "golden" datadir and make fast clones for other testing with cp -cR. The only concern is to make sure chaindata is deleted in the golden datadir, bc sometimes chaindata is already at several GBs, and bc OS blocks on copy-on-write of first byte, I've seen macos crashing and rebooting (maybe mdbx doesn't play well with copy-on-write FS with several GBs). So I just make sure it doesn't exist.