The night shift
A SysAdmin Crisis shift runs from 22:00 to 06:00 in about ten real minutes. Learn how a night works and the two ways it can end.
Last updated:
A shift is the unit of play. It begins at 22:00 and, if you can keep the fleet alive, ends at 06:00 the next morning. Everything — the score, the grade, the campaign's sense of progress — is scoped to a single night.
The clock
The simulation advances in fixed steps of 250 milliseconds, and each step moves the in-game clock forward twelve seconds. A full shift is 2400 steps, which works out to ten minutes of real time at normal pace. That ratio is fixed; difficulty changes how much happens inside those ten minutes, not how long they last.
Fixed steps are also why the game is reproducible: the same seed plays out the same way on every device. See determinism and seeds.
How a shift ends
| Ending | Condition |
|---|---|
| Sunrise | You reach 06:00 with the fleet still serving. The shift is scored and graded. |
| SLA breach | Average uptime falls below 50%. A 120-step grace period at the start keeps an early stumble from ending the night instantly. |
| Total outage | Every server in the fleet stays down at once for a full 60 seconds. |
Losing is not a soft failure. In the campaign a lost shift costs reputation, and reputation costs the next night's budget.
Uptime is the spine
Almost everything routes through a single number. Uptime falls when servers go down and when latency climbs; it also falls, separately, when an incident sits unresolved past its grace window. Since it is both the losing condition and the largest term in the score, a night spent protecting uptime is a night spent protecting the grade.
What carries between nights
Nothing about the fleet carries over; every shift rebuilds it from the night's plan. What does carry over, in the campaign, is reputation: a meta-resource that turns into extra budget the next night. See the campaign.