Looking Inside
A repository you cannot inspect is a black
box you are trusting. These are the
inspection commands, all run in the lab
against the two-snapshot repository of
chapter 6 (transcript t04).
The list of snapshots
restic snapshots
ID Time Host Tags Paths Size
c81c812b 2026-10-24 04:07:52 ... field-notes 12.000 MiB
997db8dc 2026-10-24 04:07:58 ... field-notes 20.000 MiB
Every command that takes a snapshot
accepts the short ID, a full ID, or the
word latest — the newest snapshot
matching whatever other filters you gave
(host, path, tags). latest is convenient
in conversation and dangerous in scripts
you wrote months ago; chapter 12's recipes
still use it, with their filters written
tightly enough to deserve it.
Filters worth knowing: --host,
--path, --tag narrow the list (and
what latest means) to the snapshots you
actually run a schedule for.
Walking a snapshot
restic ls latest
restic ls latest /path/inside/the/snapshot
ls prints the snapshot's tree: full
original paths, files only as names — the
contents stay in the repository until you
restore or dump them:
restic dump latest /path/to/one/file
prints a single file to standard output. For finding rather than browsing:
restic find shed-key.txt
answers with every snapshot containing a matching name and the path inside it — the command you want at the moment someone says "I deleted it on Tuesday, I think".
Comparing two snapshots
restic diff <older-id> <newer-id>
M for modified, + added, - removed,
for files and directories, ending with the
blob arithmetic of chapter 6: how much
data the newer snapshot actually added.
Diff is also the cheapest integrity habit
available — run it between last night's
snapshot and tonight's, and a backup that
suddenly "changed" 40,000 files you never
touch is a story you hear from restic
first.
How big, in which sense
restic stats # restore-size: snapshots at face value
restic stats --mode raw-data
restic stats latest # one snapshot, either mode
Chapter 6 explained the two senses:
restore-size for planning a restore,
raw-data for planning a disk. A third
mode, files-by-contents, counts unique
file contents — the number that tells you
how much of your archive is the same
photograph kept in five folders.