checkup
checkup copied to clipboard
check/http: a single error results in `down` status
I added a Pushover notifier to my checks recently and noticed just how flaky they are. By default I have a check which makes 5 attempts. If just one of the attempts fails (e.g. TLS handshake timeout) then boom the whole site is marked down, even if the other 4 work fine.
Would it be possible to consider a configuration option of allowed failed attempts?
Here's the code which configures at-least-one-error-means-down:
https://github.com/sourcegraph/checkup/blob/cdc8b7899beb72d7ffcf9b2a0e8b591add164150/check/http/http.go#L151-L157
The Result in question:
{
"title":"byparker.com",
"endpoint":"https://byparker.com",
"timestamp":1740331804280650069,
"times":[
{"rtt":5019109349,"error":"Get \"https://byparker.com\": net/http: TLS handshake timeout"},
{"rtt":149080050},
{"rtt":57264705},
{"rtt":30989184},
{"rtt":24505730}
],
"threshold":400000000,
"down":true
}