Fixing this proper will require refactoring the script generation code so it can get not only check settings, but also settings local to the probe. This work was started in #732, but still requires more effort.
Workaround
A workaround for this issue is to specify a non-empty, no-op CIDR that would filter no (real) addresses, e.g. --blocked-nets=0.0.0.0/32. This will override the default value for all checks, including multihttp.
The value of
--blocked-nets
, which defaults to10.0.0.0/8
, eventually makes it verbatim (after validation) to the--blacklist-ip
k6 flag: https://github.com/grafana/synthetic-monitoring-agent/blob/e6f47620c627339cf100d6eea50c286ec304660a/internal/k6runner/k6runner.go#L455However, specifying an empty value will not clear the in-script default value we set for multihttp checks: https://github.com/grafana/synthetic-monitoring-agent/blob/e6f47620c627339cf100d6eea50c286ec304660a/internal/prober/multihttp/script.tmpl#L43
Fixing this proper will require refactoring the script generation code so it can get not only check settings, but also settings local to the probe. This work was started in #732, but still requires more effort.
Workaround
A workaround for this issue is to specify a non-empty, no-op CIDR that would filter no (real) addresses, e.g.
--blocked-nets=0.0.0.0/32
. This will override the default value for all checks, including multihttp.