Monitors
A monitor says what to check, where to run it and when. Each run produces a result: passed or failed, how long it took, and what it found. Alert rules read those results and raise alerts.
What a monitor has
Section titled “What a monitor has”| Part | What it is |
|---|---|
| Name and Description | How it shows in lists, alerts and notifications. |
| Definition | One check (HTTP, SQL, a script, …) or a flow of several steps. |
| Schedule | A cron expression and a Time zone. |
| Runner group and Runner labels | Where it runs. Required: there is no default group. |
| Alert labels | Labels such as team=payments, copied onto its alerts. Routes and silences match on them. |
| Parameters | Optional values a run needs, such as env. |
| Client | Optional. The client it belongs to. |
| Enabled | A paused monitor doesn’t run on its schedule, and its open alerts resolve. |
Creating a monitor
Section titled “Creating a monitor”On Monitors, select New monitor. The picker has three tabs:
- Check types, grouped as Web & APIs, Databases, Logs & metrics, Code, Integrations and Several steps.
- Templates: ready-made monitors grouped as Websites & APIs, Databases, Kubernetes & Rancher, Logs, Dynatrace, Integrations, Flows & scripts, and AWS & cost. Pick one, change the address, and it runs.
- Import: a Postman collection (each request becomes an HTTP monitor, or each folder one flow) or DBeaver connections and scripts (data sources become connections, saved
.sqlscripts become SQL monitors). You see everything it would create first. Imported monitors start paused, and credentials always go into a connection, encrypted.
Schedules
Section titled “Schedules”- A schedule is a cron expression, such as
*/2 * * * *(every two minutes) or0 6 * * *(06:00 every day), read in the monitor’s time zone. - The shortest interval is 30 seconds on a server. On an instance that runs as AWS Lambda functions it is 60 seconds, and a run can start up to a minute late. On a hosted runner group, your plan sets the shortest interval.
- Each scheduled slot runs once, even with several SREInsights servers. Runs don’t overlap: the scheduler doesn’t queue a monitor’s next run while its last one is still queued or running.
Running a monitor yourself
Section titled “Running a monitor yourself”- Run now on the monitor’s page queues a run at once. If the monitor has parameters, it asks for their values first.
- A webhook trigger gives you a URL and a token. A
POSTto it starts a run, and its JSON body is available to the check astrigger.body. See Trigger a monitor from a webhook.
Neither starts a second run while one is already queued or running.
Parameters
Section titled “Parameters”A monitor can declare the values its runs need: text, a number, true/false, or one of a list. Each is required or has a default.
- Where values come from: the schedule stores a value for each parameter; Run now asks for them; a webhook call sends them in its body (
params, or top-level fields with the same names). - Where you use them:
${params.env}in a setting,params.envin an expression. A parameter also fills a connection’s${env}placeholder of the same name. Scripts find them in their input file. - Checking: SREInsights resolves and checks the values before it queues a run. A missing or wrong value is refused with
400, naming the parameter. A monitor isn’t saved while a required parameter has neither a default nor a schedule value.
Each run records the values it used and what started it.
The monitor’s page
Section titled “The monitor’s page”- Last run, Success rate and Latest value at the top.
- Recent runs: results reported by your runners. Expand a run to see what it found: flow steps, rows, series, problems.
- Series: for a PromQL check with one alert per series, each series and its state.
- Alert rules: add, edit, switch off or delete rules, with a preview against past runs.
- Open alerts and Details.
Where a run stands
Section titled “Where a run stands”| Status | Meaning |
|---|---|
| Waiting for a runner | Queued; no runner has taken it yet. |
| Passed / failed | The runner reported a result. |
timeout |
It ran past its time limit. |
skipped (no_runner) |
No runner took it in time. Fires no alert. |
error (runner_lost) |
The runner went quiet three times. |
error (dispatch_failed) |
A Lambda function could not be invoked. Counts as a failure. |
Deleting a monitor
Section titled “Deleting a monitor”Deleting a monitor stops its schedule and webhooks and resolves its open alerts. Pausing it stops its schedule and resolves its open alerts, and keeps the monitor and its history.
See also: Check types, Flow steps.