Start with the endpoint
Use the subdomain as-is for the default behavior, or add a question mark to begin custom settings.
https://flaky.unreliablewebsite.com/
Stable control page
Reliability, badly behaved.
A deliberately unreliable benchmark site for uptime monitors, synthetic checks, alert pipelines, and anyone who has ever received a 3:07 AM notification that simply said "HTTP bad".
Using params
Params are optional query string settings that let you make an unreliable endpoint behave in a specific way.
Add settings
Use name=value pairs. Join multiple settings with &. Your monitor will request the full URL.
?fail=100&status=503
Make tests repeatable
Where supported, add a seed so a random-looking test gives the same result each time.
?codes=200,500,503&seed=demo
Flaky failure
Always fail with a 503, which is useful for checking alert timing.
https://flaky.unreliablewebsite.com/?fail=100&status=503
1 / 6
- Always fail with a 503, which is useful for checking alert timing.
- Force a 200 OK response so you can confirm the monitor is configured correctly.
- Wait eight seconds before responding, useful for testing timeout thresholds.
- Fail for the first fifteen seconds of every minute, then recover.
- Return 200 OK without the expected keyword, useful for content monitoring.
- Use a seed so the same random-style endpoint gives a stable benchmark result.
Test cases
Each endpoint is small, configurable, and intentionally annoying in one specific way.
LAB-01
HTTP
Random Status
Picks from a configured set of status codes. Sometimes a teapot, sometimes a disaster.
- Subdomain
random-status.unreliablewebsite.com- Assert
- Alert on non-2xx or compare chosen status distribution.
Params: codes=200,500,503, seed=abc, body=1
LAB-02
FLAKE
Flaky
Fails by percentage, for teams who miss their incident channel being too quiet.
- Subdomain
flaky.unreliablewebsite.com- Assert
- Detect repeated intermittent 503s without declaring total doom.
Params: fail=25, status=503, seed=abc
LAB-03
LATENCY
Slow
Responds eventually, after taking a meaningful pause to consider its choices.
- Subdomain
slow.unreliablewebsite.com- Assert
- Alert when response time crosses your monitor timeout.
Params: delay=5, min=1, max=10, status=200
LAB-04
CLOCK
Scheduled
Fails inside predictable windows so incident timing can be compared fairly.
- Subdomain
scheduled.unreliablewebsite.com- Assert
- Detect the down window and recovery time accurately.
Params: down_for=5, period=60, status=503
LAB-05
LOOP
Redirect Loop
Redirects in circles until the monitor admits it is lost.
- Subdomain
redirect-loop.unreliablewebsite.com- Assert
- Detect redirect loops and excessive redirect chains.
Params: limit=10, status=302
LAB-06
BODY
Wrong Content
Returns 200 OK with the wrong words, which is technically fine and practically not.
- Subdomain
wrong-content.unreliablewebsite.com- Assert
- Use keyword checks, not just status checks.
Params: expected=..., mode=wrong|right
LAB-07
JSON
Bad JSON
Advertises JSON and then forgets where the brackets go.
- Subdomain
bad-json.unreliablewebsite.com- Assert
- Validate API payload shape, not just content type.
Params: mode=bad|good, status=200
LAB-08
METHOD
Method Mismatch
GET and HEAD disagree, as many monitoring setups quietly discover.
- Subdomain
method-mismatch.unreliablewebsite.com- Assert
- Confirm which HTTP method your monitor actually uses.
Params: head=500, get=200
LAB-09
LIMIT
Rate Limit
Starts friendly, then returns 429 when the visitor gets clingy.
- Subdomain
rate-limit.unreliablewebsite.com- Assert
- Check handling of monitor frequency and 429 responses.
Params: limit=5, window=60, status=429
LAB-10
HEADER
Headers
Sends odd cache and length hints for header-sensitive monitors.
- Subdomain
headers.unreliablewebsite.com- Assert
- Notice stale cache headers and malformed response metadata.
Params: cache=stale, length=wrong|omit|normal, status=200