Expectation monitoring explained
What is expectation monitoring?
Expectation monitoring checks whether recurring automated work produced the expected result, not only whether a process ran. In Glnc, a Check combines an interval or cron schedule with an IANA timezone and optional output rules. Runs report status and metadata; missing, late, failed, slow, or unmet output becomes visible status and response context.
What is an expectation?
An expectation is a durable statement that recurring work should happen on a cadence and produce a particular result. A Glnc Check records that statement. It can use a rolling interval or a calendar schedule expressed as cron plus an IANA timezone.
The automation continues to run in its existing scheduler, service, or agent. It reports each Run to Glnc through the REST API.
What makes an expectation unhealthy?
A Check can become unhealthy when a Run is missing or late, reports a failed or timed-out status, takes longer than a configured duration budget, or fails an output rule. The resulting status keeps timing and reported evidence together for review.
How do output expectations work?
A Run can include a metadata object such as { "records_created": 42 }. A latest-run rule evaluates the newest value. A windowed rule can evaluate an average, sum, minimum, maximum, or count across a trailing window.
Each rule names a bad-state condition and fires when that condition holds: a warning rule produces a warning state, a critical rule produces a failing state. A succeeded Run therefore reports execution status, but it does not by itself prove that no output rule tripped.
What happens when an expectation fails?
Glnc presents the status reason with the Check’s runner, affected-system labels, and remediation note. An acknowledgment keeps the unhealthy status visible while suppressing immediate alerts until recovery or expiration.
A remediation work item is optional. Glnc can open one only when the Check has a remediation queue and stage configured.
When is expectation monitoring a fit?
Use it when someone currently checks a business or operational result after recurring work: imported records, generated artifacts, completed handoffs, or another measurable output. If silence is the only failure signal you need, a simpler heartbeat may be enough.
Next step
Define one consequential expectation.
Read the API contract before reporting production evidence, or join the private beta to start with one recurring automation.