On April 12, 2026, at about 2 a.m. UTC, scheduled functions on several Netlify sites stopped firing. Nobody had deployed or changed a setting, and the Functions page kept promising “next execution in 10 minutes.” The forum thread filled with “same here” until things recovered about two days later. That’s the hard version of this problem. Most cases are easier, and they’re listed below in the order we’d check them.
First, Did It Actually Not Run?
Open Functions in the Netlify UI, select the function, and look at its logs. Two things make an empty log misleading. A Netlify support engineer pointed out that a scheduled function only shows in the logs if it logs something, so add a console.log first. And check the time range: one forum user’s runs only appeared after switching the view from “latest” to “last hour”.
Then click Run now. If the manual run works, your code is fine and the problem is the trigger. If it fails, you have an ordinary bug, and the schedule is a red herring.
Six Reasons a Netlify Scheduled Function Isn’t Running
1. You’re Looking at a Preview, Not the Published Deploy
Netlify’s scheduled functions documentation is explicit: scheduled functions “only run on their schedule for published deploys.” Deploy Previews and branch deploys never fire on their own. The same applies to a new schedule. If you’ve locked publishing or stopped auto-publishing, your latest deploy isn’t the published one, and the schedule that’s actually running is whatever the published deploy carries.
2. The Schedule Is Declared Somewhere Netlify Doesn’t Read It
There are two places to put a schedule: an exported config object in the function file, or a [functions."name"] block in netlify.toml. Inline schedules only work for TypeScript and JavaScript. Any other language must use netlify.toml.
Framework builds add a wrinkle. In a long-running forum thread, several people with syntactically correct inline schedules only got them registered by moving the schedule into netlify.toml, and another found scheduled functions inside an existing Next.js project never ran while the same functions worked in a standalone repo. When in doubt, the toml file is the declaration least likely to be lost in a build step. A correctly registered function shows a Scheduled badge on the Functions page; no badge, no schedule.
3. It Ran, Just Not When You Expected
Schedules run in UTC. 0 9 * * * is 9 a.m. in London for part of the year and 4 a.m. in New York. The Functions page converts the next execution time to your own timezone, which helps, but the expression itself is always UTC. Netlify also supports @hourly, @daily, @weekly, @monthly and @yearly, though not @reboot or @annually. And @weekly means Sunday at 00:00 UTC.
4. It Started and Was Cut Off at 30 Seconds
Scheduled functions have a 30-second execution limit. A job that usually finishes in 20 seconds and occasionally needs 40 will fail on exactly the days it has the most work. Netlify points longer tasks to background functions. If the job is close to the limit, split it or move it, because the cut-off is not something you can tune away.
5. You’re Trying to Trigger It With a URL
You can’t invoke a scheduled function directly with a URL in production. Local URL invocations under Netlify Dev are for debugging only, and Netlify Dev never runs the schedule at all. To test, use Run now in the UI or netlify functions:invoke <name> locally. Scheduled functions also ignore request payloads, so anything that expects a POST body belongs in a regular function.
6. Netlify’s Scheduler Is Having a Bad Day
This is the April 12 case, and a May 2026 thread describes the same symptom: Scheduled badge present, correct next-execution times, Run now working, and no scheduled runs. Nothing in your repository can fix that, and nothing in the Netlify dashboard will tell you it’s happening. It’s the reason the next section exists.
The Dashboard Can’t Tell You a Run Didn’t Happen
Every cause above has one thing in common: the failure produces no event. No error, no failed invocation, no log line. A countdown to the next execution is a statement about the schedule, not about whether the last run happened.
The only reliable signal for “didn’t run” is something outside Netlify that expects to hear from your function and complains when it doesn’t. That’s a heartbeat, also called a cron monitor or dead man’s switch. We covered how it differs from an uptime check in heartbeat vs cron vs uptime monitoring, but the short form is: your job pings a URL when it finishes, and silence past the expected time raises an alert.
In FlareWarden, add a cron monitor under the site’s uptime monitor, set the schedule, and copy its ping URL. FlareWarden reads cron expressions in UTC too, so you can paste the same expression you gave Netlify. Then ping it from the function:
// netlify/functions/nightly-sync.mts
import type { Config } from "@netlify/functions";
const PING = "https://app.flarewarden.com/ping/YOUR-TOKEN";
const ping = (path = "") =>
fetch(`${PING}${path}`, { signal: AbortSignal.timeout(5000) }).catch(() => {});
export default async () => {
await ping("/start");
try {
await doTheSync();
await ping("/complete");
} catch (err) {
await ping("/fail");
throw err;
}
};
export const config: Config = { schedule: "0 3 * * *" };
Each failure mode now shows up as something specific. A missed trigger (causes 1, 2, 5 and 6) means no ping, so the monitor goes Late and alerts. An exception sends /fail. A run killed at 30 seconds sends /start but never /complete; set the cron monitor’s max run duration to 1 minute (its shortest setting) and that becomes a hung-job alert instead of silence. The cron monitor documentation covers the states and the /start, /complete and /fail signals in full.
Two practical notes. Keep the ping’s own timeout short, since it spends the same 30 seconds as your job. And pick a grace period that matches how late Netlify’s runs normally land for you, so that a run a few minutes late doesn’t page anyone.
Which Jobs Deserve This
Not every scheduled function needs a monitor. A function that warms a cache every hour recovers on the next run, and an alert for it is noise. Monitor the ones whose silence costs something: syncs that feed other systems, billing and reminder emails, and anything that writes database backups or exports. The pattern is the same one we described for Vercel cron jobs that don’t run, because the blind spot is the same: a platform scheduler reports what it intends to do, not what it did.
Key Takeaways
- Published deploys only. Deploy Previews and branch deploys never run on their schedule, and neither does an unpublished newer deploy.
- Check where the schedule lives. Inline
configworks for TypeScript and JavaScript; everything else, and the safer choice for framework builds, isnetlify.toml. - Schedules are UTC, runs stop at 30 seconds. Most “it never ran” reports are really “it ran at a different hour” or “it got cut off.”
- The countdown isn’t evidence. In April 2026 the dashboard showed next-execution times for two days while nothing fired.
- Heartbeat the jobs that matter. A ping on completion plus a max run duration turns every silent failure into a specific alert.
Want to know the night a Netlify job doesn’t run? Start monitoring free with FlareWarden: cron and heartbeat monitors are included on the free plan, alongside 15 monitors and no credit card. See how cron monitoring works.