Glossary
Short definitions of the words that come up when you run production systems: what each one means, why it matters, and where on this site it is covered in depth. Written to be read in thirty seconds.
Checks, intervals, thresholds, certificates and domains.
cron monitoring, dead man's switch monitoring
Heartbeat monitoring is a technique where a job pings a unique URL upon completion, allowing the monitor to detect if the job failed to run or hung.
Uptime monitoring checks a website, API, port or hostname from outside on a fixed interval and records whether it answered correctly and how long it took.
Alerts, escalation, rosters, rotations and the noise problem.
Alert fatigue is the state where people receive so many alerts, or so many that need no action, that they stop reacting to them, including the ones that matter.
escalation chain
Escalation policy is an ordered list of levels that defines who is notified about an alert, through which channels, and how long to wait for an acknowledgement before moving to the next level.
Follow the sun is an on-call model where responsibility moves between teams in different timezones so that nobody is paged at night.
On-call is the arrangement where a named person responds to service alerts during a defined period, including outside working hours.
shift swap
On-call override: a dated exception to the schedule where a different person covers a window than the rotation produces.
on-call rota, call roster, duty roster
On-call roster: ordered list of people taking turns being on call for a service, plus the rotation rule deciding whose turn it is.
On-call rotation is the rule that advances shifts through a roster, setting how long each shift lasts, when handover happens and who is next.
On-call schedule: the calendar of who is on call on which dates, produced by applying a rotation rule to a roster.