2023-12-03

ZFS snapshots 

identifier definition:  pool/dataset
where 'dataset' could be a zfs filesystem, volume [block-like], 'snapshot' or 'bookmark'
and 'pool' is the zfs unit doing the storage management for different datasets

example:   home/nt  (pool:home; dataset:nt, which is a filesystem, which just happens to be mounted on /home/nt )


to make a snapshot of a particular dataset:
note: this is only possible at dataset level, not lower levels (subdirectory)
 # zfs snapshot  pool/dataset@arbitraryname


examples
 # zfs snapshot home/nt@$(date +%F_%H%M%S) 
 # zfs snapshot home/public@some_name_for_a_snapshot 

list snapshots 
 # zfs list -t snapshot 

remove a snapshot by name
note that if there are things depending on the snapshot, it won't actually be 
removed until they've gone ... there's a switch '-d' to be firmer about it 
 # zfs destroy home/nt@2023-12-02_224947

more thorough matches can be made by putting snapshot identifiers in a 
comma-separated list, and by making ranges with '%' between, or open ranges
by omitting one side of the '%'; and ranges can be included in a list, e.g.
 # zfs destroy home/srv@thirdsnapshot,20220812%20221231,20230923%  


when a snapshot exists, it can be accessed through ".zfs/snapshot/snapshotname" 
within the directory where the dataset (zfs filesystem) is mounted, e.g.:
 # ls /home/nt/.zfs/snapshot/2023-12-02_234315/
note!!!! this .zfs subdirectory doesn't show in listings, but it appears when
one tries to access things under it: have faith as you type 

more detailed listings with optional content and sorting 
 # zfs list -t snapshot -o name,creation,logbias,quota,primarymount -s creation
 # zfs list -t all  -o name,creation,logbias,quota,primarycache -s quota 

checking manual pages, e.g.
 # man zfs-list 
 # man zfs-destroy   
i.e.: command and its mode, hyphenated