On Checkly?

Checkly synthetic checks flaking or heartbeats silent? We fix it.

Checkly makes a critical user journey into a scheduled Playwright script and a real API call into a repeatable probe. When it works, the first alert you get is from your own synthetic rather than a customer. When it does not, the checks either flake constantly or stop running because the runtime budget was quietly spent.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

checkly-run.log

still failing

What it is

Checkly at a glance

Checkly is a synthetic monitoring and API checking platform founded in Berlin in 2018 by Hannes Lenke and Timo Euteneuer. It runs scheduled checks from global locations, including real browser flows written as Playwright scripts, HTTP API assertions, and heartbeat monitors that your own jobs can ping.

The product leans into monitoring as code. Checks are defined in a TypeScript or JavaScript project that lives in your repository, which means pull requests can review the monitors the same way they review the application. Checkly also offers a Terraform provider and a CLI, so teams managing infrastructure as code can keep synthetic monitoring in the same workflow.

Type
Synthetic monitoring, API checks and heartbeats
First released
2018
License
Proprietary SaaS, free tier and usage based pricing
Maintained by
Checkly GmbH, Berlin
Typically used for
Scheduled Playwright user journey checks, API endpoint monitoring, heartbeat pings, visual regression

The honest version

Checkly pros and cons

No technology is all upside. This is where Checkly earns its place and where it tends to cause trouble.

Where Checkly is strong

  • Monitoring as code is a real improvementWriting checks as a TypeScript project kept in version control means monitoring reviews, rollback and reuse look like code review rather than clicking through a dashboard.
  • Playwright under the hoodSynthetic checks run the same Playwright that your end to end tests already use, so the author experience and the debugging tools are familiar.
  • Global check locationsChecks run from a long list of regions, and the differences between them are often what tell you a CDN or routing problem is region specific.
  • Heartbeat pings for batch jobsCron jobs and workers that need to be noticed when they stop running are a first class check type, not a workaround.
  • Terraform and CLIThe provider and the checkly deploy workflow make it reasonable to manage dozens or hundreds of checks without every change going through the UI.
  • Reasonable free tierEnough runs for a small team or a side project, and the pricing scales in units that are easier to reason about than per metric observability bills.

Where Checkly bites

  • Runtime quotas are easy to blowA check set to run every minute from five regions uses 7,200 runs per day for one check. Teams underestimate that number and find the budget exhausted midway through a month.
  • Flaky checks erode trust fastA Playwright flow against a real third party login or payment page will occasionally fail for reasons that have nothing to do with your service, and the team learns to ignore the alert if it is not tuned down.
  • Private location agents need upkeepRunning checks from inside a private network requires a long lived agent, and when it goes offline the checks from that location silently stop running.
  • Visual regression is noisy by defaultPixel level diffs trigger on fonts, ad slots, personalised content and anything else that moves, so the visual check feature needs careful allow listing to be useful.
  • Runtime version upgrades are mandatoryCheckly deprecates older Playwright runtimes on a cadence, and a check that was passing last quarter can start failing because the runtime is no longer supported.
  • Smaller integration ecosystem than DatadogAlert channels and incident platforms are covered, but deep integrations into APM, log search and broader observability still need glue.

Sound familiar?

Common Checkly problems we fix

The Checkly problems that bring people to us, and what you have once each one is fixed.

run cost: 184 check runs this month, quota 200

Runtime budget is running out

A high frequency check from multiple regions quietly consumes the monthly quota, and once it is spent the checks stop running rather than overrun the bill. Cadence and region tuning usually recovers most of it.

Checks fit inside the plan comfortably

heartbeat missed: no ping received in 2h

A heartbeat is not being pinged

The batch job is failing earlier than the ping, or a deploy renamed the heartbeat and the old URL is still in the code. We trace the first failing step rather than only extending the heartbeat window.

Heartbeats honest about the job

page.click Timeout 30000ms exceeded

Synthetic Playwright check is flaky

A locator that works in your own test runs is racing with an animation, a cookie banner, or a third party widget on the live site. We make the selector more specific and add the right waits.

Checks stable across runs

runtime not supported: Playwright 1.40

Playwright runtime is deprecated

Checkly retired the runtime version your checks were pinned to. Upgrading usually needs a handful of API changes, which we work through in one pass rather than per check.

Checks on a supported runtime

private location agent offline

Private location agent is down

The agent lost its connection to Checkly, probably because the host restarted or network egress changed. We restart it, pin the version, and set up a heartbeat on the agent itself.

Private checks running again

ERR_CERT_DATE_INVALID

SSL certificate expiry caught by a check

A certificate near expiry trips a browser error that your production might be routing around with warnings. The check caught it before customers did; the fix is on the certificate side.

Certificates healthy, checks green

snapshot mismatch: homepage.hero

Visual regression on the homepage

A new campaign image, a font swap, or a layout shift that nobody flagged. Sometimes a real regression, sometimes a planned change the visual check was never told about.

Visual baseline matches intent

Terraform error: conflicting schedule and tags

Monitoring as code is drifting

A manual change in the UI diverged from the Terraform state, and the next plan reports a conflict. We bring both back into sync and lock down UI changes where it matters.

State and reality aligned

What we do

Checkly help, from one flaky check to monitoring as code

Call us for one unreliable check, or let us design the whole synthetic setup.

Rescue
Flaky synthetic checks, missing heartbeats or exhausted runtime budgets. We find the real cause and stabilise before anything else.
Synthetic check design
Playwright flows against the journeys that actually matter to revenue, with selectors, waits and assertions that survive real site behaviour.
API check coverage
Endpoint checks with status, latency, response body assertions and signed requests, scheduled from the regions your customers actually come from.
Monitoring as code
Checks moved into a TypeScript project in your repository, with pull request review, CLI deploys and Terraform where it fits.
Private location setup
Agents installed inside private networks, verified against the same checks running from public regions, and monitored so a dead agent is noticed.
Managed Tech
Monthly reviews of flake rate, runtime usage, visual regression noise and runtime version support, plus a hand on deck when a live check starts failing during a launch.

How it works

From first message to fixed

  1. Look Free estimate

    Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.

  2. Fix

    We fix the cause, not the symptom. If something is down, getting it back up comes first.

  3. Harden

    Backups, monitoring, updates and access, so the same problem can't quietly come back.

  4. Hand back or stay on

    You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.

Often found together

Running Checkly alongside something else?

We work on the whole stack, so the problem can be wherever it actually is.

Questions

Checkly questions

Something else? Call+1 737 727 7327

Why is our Checkly runtime budget exhausted so quickly?

The common cause is a check running every minute from four or five regions, which is 7,200 runs per day for one check alone. Dropping to five minute intervals outside of business hours, or picking two representative regions rather than all of them, usually recovers most of the budget without losing coverage on the paths that matter.

How do we stop synthetic checks from being flaky?

Flakes usually trace back to locators that are too broad, missing waits for animations or third party scripts, and shared accounts being used by multiple checks at once. We make the selectors more specific, use Checkly's double check feature to filter transient failures, and give the flow its own test account where it makes sense.

Why did a heartbeat miss when the job clearly ran?

Three usual causes: the job succeeded but the ping step was placed after a condition that short circuits, the URL was changed on the Checkly side without updating the job, or the job is finishing just after the heartbeat window. We inspect the job, align the schedule, and add a retry on the ping itself.

Should we use Checkly or Datadog Synthetics?

Checkly wins on monitoring as code, Playwright script familiarity and friendly pricing for teams that are not already in Datadog. Datadog Synthetics wins when you want checks to sit alongside APM, logs and dashboards in one place. We benchmark both against your real use cases before suggesting one.

Can you help move our checks into code?

Yes. We export the existing checks, rewrite them as a TypeScript project with the Checkly CLI, add continuous deployment from your repository, and verify behaviour matches the old dashboard checks before anyone turns the UI editing off.

How much does Checkly help cost?

The quote is free. Send over a failing check or the Checkly project, and we come back with a price and a short plan before any changes land.

Free estimate

Checkly letting your team down?

Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.

  • No charge to find out what it will take
  • If it's urgent, say so

Replies come from a person. By sending, you authorize us to email you back. No marketing.