You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* benchmark presence checking on both a map and a slice
* refactor metric processing
* remove `iterations` metric
Not seen in practice with iterations=1, but part of the test suite
* create `http_status_code` metric
* keep `status` label only on `http_requests_total` metric
* remove high-cardinality `error` label
* create `http_version` metric
* remove most of the output unit tests
* generate http_duration_seconds{phase="resolve"} hardcoded to zero
* use custom sm phase names for http_duration_seconds
* make `script_duration_seconds` actually seconds
* integration: test for custom value of `tls_version` label
* strip `tls` prefix of `tls_version` label value
* integration: run subtests in parallel
* do not strip `status` tag from any metric
* Fix typo in output test
Co-authored-by: Dougal Matthews <[email protected]>
* keep `tls_version` and `proto` labels on all metrics
To be consistent with the `status` label, and to ensure that we do not
cause data loss due to duplicates. This should not impact cardinality.
* make `probe_http_requests_failed_total` a counter instead of a rate
`probe_http_requests_failed` is emitted as a rate now, as the original
metric is
---------
Co-authored-by: Dougal Matthews <[email protected]>
0 commit comments