Your First Sync

This chapter runs the array for the first time: the first sync, a look at status, and the diff habit that should precede every later sync. Everything here was run in the lab; transcripts t02–t04 are the full output.

Before: a "before" picture

The lab recorded checksums of every file before touching SnapRAID (transcript t01). Do the equivalent on a real array in spirit: know roughly what is on your disks and how full they are. The checksums are what later chapters verify recoveries against — recovery claims should always end with a check, not a feeling.

sync — bring the parity up to date

$ snapraid -c snapraid.conf sync
...
Everything OK

The first sync reads every file, computes every block's hash, writes the full parity file and the content files, then verifies what it wrote. On the lab's 24 MB it took seconds; on several terabytes of real media it takes hours, once. Later syncs only touch what changed.

Check: the parity file exists and is about the size of the fullest data disk's data:

$ ls -la parity/ disk1/snapraid.content
-rw------- 1 ... 13631488 ... snapraid.parity
-rw------- 1 ...     1931 ... snapraid.content

(The lab's data disks held about 13.6 MB between them; parity tracks the larger disk's usage, block for block.)

status — the array at a glance

$ snapraid -c snapraid.conf status
SnapRAID status report:

   Files Fragmented Excess  Wasted  Used    Free  Use Name
       3       0       0    -0.0       0     203   0% d1
       4       0       0     0.0       0     203   0% d2

Below the table, status draws the scrub-age graph (chapter 6) and ends with the lines worth reading every time:

No sync is in progress.
100% of the array is not scrubbed.
No error detected.

"No error detected" means no command has found an error so far — it is a report of the array's record, not a fresh proof. Fresh proof is check's job (chapter 6).

diff — the pre-flight check

diff compares the disks against the content file and lists what a sync would do, without doing it:

$ snapraid -c snapraid.conf diff
update music/new-track.flac.bin

       6 equal
       0 added
       0 removed
       1 updated
       0 moved
       0 copied
       0 relocated
       0 restored
There are differences!

Exit code 0 means no differences; 2 means differences were found (transcript t04). Read the categories before every sync:

• removed is the dangerous one. A few removals you made on purpose are fine. A disk's worth of removals means a disk is missing, unmounted, or empty — do not sync; find out why first. Sync makes the current state the protected state, and chapter 7 shows what that does to a file deleted by accident.
• added / updated are the normal churn of a living array.
• moved / copied are recognised and cost no parity work.

The rhythm to learn

snapraid -c snapraid.conf diff     # look first
snapraid -c snapraid.conf sync     # then commit
snapraid -c snapraid.conf status   # confirm

The lab's second sync (t04) shows the steady state: one new file, a short sync, "Everything OK". Chapter 9 turns this rhythm into a schedule, with one protection added: a limit on how many deletions a scheduled sync may accept before it refuses and asks a human.

The one-sentence version

sync makes the parity match the disks as they are right now, diff shows you what that would mean before you commit to it, and status is the report card you read afterwards.