Every machine, every job, every failure — on one screen.
Jobs stream live as they run: the console is fed by the backend as events happen, so what you are looking at is the current state and not a page you have to remember to refresh. No per-device remote session to find out whether a backup worked.
Thirty days, forty-one machines, one strip.
A health score over the last 30 days, how many devices are green today, how many failures there were this week, and how many agents have gone quiet. The score is a number you can be held to, not a mood.
Silence is the real failure
Agents check in about once a minute, and a device that goes quiet is emailed to the workspace owners and admins rather than left in a queue for someone to notice. The failure mode of most backup setups is not a loud error; it is months of quiet.
A queue of what is broken
Failed snapshots are listed oldest first with the cause spelled out — offline with no heartbeat for two days, target volume full with zero bytes free, three scheduled runs missed because the machine was off.
Recovery readiness, not just success rate
How many devices actually backed up in the last seven days, which is a different question from how many jobs succeeded. Two machines with no recent backup are two machines, however green the rest of the fleet looks.
Forty-one machines, and what each of them last did.
Registered devices with their policy, their status, when they last backed up and how much is protected. Two need attention and one is offline — and the list says which, rather than leaving it to be worked out.
| Device | Policy | Status | Last backup | Protected |
|---|---|---|---|---|
| ACC-SRV-01Windows Server 2022 · agent 2.4.1 | Nightly | Online | Running now · 36% | 1.9 TB |
| WH-NAS-02Windows Server 2019 · agent 2.4.1 | Shared drives | Online | Running now · 13% | 842 GB |
| FIN-LAP-14Windows 11 Pro · agent 2.4.1 | Laptops | Online | Today 09:12 · verified | 184 GB |
| FIN-LAP-09Windows 11 Pro · agent 2.4.1 | Laptops | Online | 4 days ago · 3 runs missed | 151 GB |
| DEV-WS-03Windows 11 Pro · agent 2.4.1 | Nightly | Online | Yesterday 02:14 · failed | 410 GB |
| RETAIL-POS-07Windows 10 IoT · agent 2.3.9 | Retail | Offline | 2 days ago · completed | 28 GB |
| OPS-WS-11Windows 11 Pro · agent 2.4.1 | Laptops | Online | Today 08:40 · verified | 96 GB |
The event stream, in the order it happened.
Backups started and finished, restores completed, failures with their cause, missed slots with their reason, and devices that stopped calling home. Seven days by default, exportable as a log.
Missed slots are not retroactively caught up. The next scheduled run is the next one — we would rather tell you that than quietly stack six overdue jobs onto a laptop at 9 am. What the sweeper gives you instead is the distinction: a machine that was switched off looks different from an agent that failed to fire.
Six answers, and only three of them need you.
Every run ends in one of these, and the console never rounds a partial outcome up to a success. “It ran” and “it got everything” are two different answers, so they are two different statuses.
| Job status | What it means | Does it need you |
|---|---|---|
| pending · scanning · running · verifying | The run is alive. The console streams the percentage as it moves. | Nothing. |
| completed | Every source path was stored and the snapshot was checked against the repository afterwards. | Nothing. |
| completed_with_errors | One source path failed. The rest succeeded, and the failure is named. | Read the named path. Usually a permission or a removed drive. |
| completed_unverified | The data was written, but the integrity check did not finish. It ran; we cannot yet say it got everything. | Treat that point as unproven until the next run verifies. |
| failed · cancelled · interrupted | No backup was produced. Interrupted usually means the machine slept or lost power mid-run. | Yes. Nothing was stored for that slot. |
| skipped | The slot never started, and carries a reason: device_offline, agent_not_reporting, previous_run_active or backups_paused. | The first two need you. The other two do not. |
Workspaces, roles and expiring invites
Separate workspaces per client or per site, owner / admin / member roles governing who can administer a workspace and who can remove things from it, and invite links that expire rather than sitting in an inbox forever.
Tenant isolation by construction
Every workspace is a separate bucket and a separate set of devices, and the workspace on a request is taken from the verified session token — never from a header, a body field or a URL.
Logs pulled without a remote session
When a run misbehaves, the console can ask that device for its recent logs. The agent zips and uploads them itself, while the failing backup is still running.
About running a fleet from one screen.
Do I have to refresh the console to see a job move?
No. Jobs stream live: the console is fed by the backend as events happen, so what is on the screen is the current state. There is no refresh button and no per-device remote session needed to find out whether a backup worked.
How do I find out a backup stopped happening?
The console tells you, and so does your inbox. Agents check in about once a minute and a sweeper walks every schedule slot using the same arithmetic the dashboard predicts with. Failed runs, missed schedules, a device that stopped calling home and storage crossing its limit are eighteen kinds of alert between them.
What is the difference between a missed run and a failed one?
A failed run started and produced nothing. A missed slot never started, and carries a reason: device_offline, agent_not_reporting, previous_run_active or backups_paused. The first two need somebody; the other two do not. That distinction is the whole point of the sweeper.
Can I keep clients or sites separate?
Yes. A workspace is a separate bucket and a separate set of devices, with its own roles and invites. The workspace on any request is taken from the verified session token rather than from anything the request itself states.
Is the console data real?
The device names, job figures and dates shown on these pages are demo data, as the live console labels them. The console itself is at backup.comsky.ai.
Open the console → Every device name and figure on this page is demo data.
Start with Comsky Backup.
Grow into the ecosystem.
One account opens every Comsky product — and one invoice covers them, however many you run.