Alert Rules

Alert rules let you fire a silent failure alert based on the output of a successful job — without your job needing to throw an error.

The problem they solve

Some jobs succeed technically but produce a bad outcome. A database sync job that connects fine but processes zero rows is a success from the process's perspective, but it probably shouldn't be silent.

With alert rules, you can configure Crontify to fire an alert whenever a job succeeds but the result doesn't look right.

How they work

1. Your job sends a /success ping and includes a meta object with numeric fields
2. Crontify evaluates each alert rule against the meta object
3. If a rule's condition is true, a silent_failure alert fires on all attached channels
4. The run retains success status — a silent failure is not a crash

Rule structure

Each rule has three parts:

Field	Description
Field	A key from the meta object — alphanumeric and underscores, max 50 characters
Operator	One of lt, gt, eq, ne
Threshold	A numeric value to compare against

Operators

Operator	Meaning	Example
lt	Less than	rows_processed lt 100 — alert if fewer than 100 rows
gt	Greater than	error_count gt 0 — alert if any errors occurred
eq	Equals	rows_processed eq 0 — alert if exactly zero rows
ne	Not equals	exit_code ne 0 — alert if exit code is non-zero

A rule fires when the condition evaluates to true. rows_processed eq 0 fires when the value is zero — not when it fails to be zero.

What happens if the field is missing?

If the field named in a rule is absent from the meta payload, the rule is silently skipped — no alert fires. A missing field and a zero value are treated as different things by design.

SDK usage

await monitor.success({
  meta: {
    rows_processed: 0,      // A rule of `rows_processed eq 0` would fire
    emails_sent: 0,
    duration_ms: 1200,
  },
});

await monitor.fail({
  message: 'Query timed out',
  meta: { exit_code: 1 },
});

Configuring rules

Rules are configured per monitor from the monitor form (New Monitor or Edit Monitor). Each monitor supports up to 5 rules.

To add a rule:

1. Scroll to the Alert Rules section of the monitor form
2. Enter a field name, choose an operator, and set a threshold
3. Click + Add Rule
4. Save the monitor

Rules take effect immediately — the next success ping that includes the relevant field will be evaluated.

Alert deduplication

Silent failure alerts follow the same cooldown rules as other alert types. If your job fires a silent_failure alert every minute, the cooldown (default 60 minutes) ensures you're not flooded with notifications.