Surface k6 log output when k6 exits with an error

[nadiamoe]

Currently, the local k6 runner will return an error if the k6 binary exits with an error:

synthetic-monitoring-agent/internal/k6runner/local.go

Line 123 in 4c44fa9

return nil, fmt.Errorf("cannot execute k6 script: %w", err)

Unfortunately, it does so before collecting logs a few lines later:

synthetic-monitoring-agent/internal/k6runner/local.go

Line 139 in 4c44fa9

result.Logs, err = afs.ReadFile(logsFn)

Furthermore, the k6 processor will ignore any reported logs if k6runner.Run returns an error, as it is currently doing:

synthetic-monitoring-agent/internal/k6runner/k6runner.go

Line 130 in 4c44fa9

result, err := k6runner.Run(ctx, r.script)

This makes k6 log output invisible when there is an error running k6, which results in hard to debug scenarios. We should change this:

• Always return k6 logs, if present
• Report errors occurred after running k6 as part of RunResponse.Error and RunResponse.ErrorCode, instead of error in Run():
  

  synthetic-monitoring-agent/internal/k6runner/k6runner.go

  Line 75 in 4c44fa9

  Run(ctx context.Context, script Script) (*RunResponse, error)

The cloud runner does this by heuristically checking if an error is an user-error, and reading the log output if it is:
https://github.com/grafana/sm-k6-runner/blob/f0057a93a7237b827aca0d1dc470cd45df0f79f3/internal/runner/runner.go#L213

[peterschretlen]

Related to #1059 ?