Job statuses
Every backup run ends in exactly one status. The Activity screen groups them into five buckets, and the counts beside each filter add up to the total because every status belongs to one bucket and no other.
In flight
Nothing to act on yet — the run has not produced an outcome.
| Status | Meaning |
|---|---|
pending |
Queued; the agent has not started it |
scanning |
Working out what changed since the last run |
running |
Reading and uploading data |
verifying |
Checking the integrity of what was just written |
Completed
| Status | Bucket | Meaning |
|---|---|---|
completed |
Success | Everything asked for made it into the recovery point |
completed_with_errors |
Warning | A source path failed. The rest of the run succeeded |
completed_unverified |
Warning | The integrity check on the written snapshot did not pass |
Both warning statuses produce a usable backup, and both are shown amber rather than green.
completed_unverified is the one to understand. Every snapshot gets an
integrity check after it is written. When that check fails, the agent records
completed_unverified along with the reason. The snapshot exists and may well
restore perfectly — but a recovery point that could not be verified is not a
clean one, and it is not counted as a success.
Did not produce a backup
| Status | Meaning |
|---|---|
failed |
The run started and something broke on that machine |
cancelled |
Stopped deliberately |
interrupted |
Ended without finishing — for example the machine slept or lost power |
Never started
skipped is its own bucket, deliberately separate from failures. A failed run
tried and something broke; a skipped run never happened, and for two of the
four reasons that was the correct outcome.
| Reason | Act on it? |
|---|---|
device_offline |
Yes — the machine was not reachable at the scheduled time |
agent_not_reporting |
Yes — the agent has stopped checking in |
previous_run_active |
No — the previous run was still going |
backups_paused |
No — backups were paused deliberately |
Counting skipped runs as failures would make a healthy policy read as broken, and would bury the two cases worth acting on inside the same red number.