Summary

Cypress test scripts that chain multiple cypress run invocations with && do not execute the second (or subsequent) runs when the first run fails. This leads to incomplete test coverage in CI and misleading pass/fail results.

Description

In frontend/package.json, several npm scripts run multiple Cypress suites in sequence using the shell operator &&. If the first cypress run exits with a non-zero status (e.g. one or more tests fail), the shell short-circuits and all following cypress run commands are never executed.

• Current behavior: First run fails → later runs are skipped → only the first suite's results are visible; overall run may still be reported as failed, but later suites are never run or reported.
• Expected behavior: All configured suites run regardless of individual pass/fail, and results (and JUnit reports, where used) reflect the full set of runs so that failures are visible per suite and CI can report complete status.

Affected scripts (kiali/kiali and kiali/openshift-servicemesh-plugin)

The following scripts are affected in frontend/package.json (and equivalent structure in the plugin repo where present):

In each case, a failure in the first run prevents the second (and any further) run from executing.

Affected repositories and branches

• Repositories: kiali/kiali, kiali/openshift-servicemesh-plugin
• Branches: All release branches and main/master where these scripts exist (same package.json script definitions).

Suggested direction (for implementation)

• Run all suites in a single pass where possible (e.g. single cypress run with combined tags), or
• Use a runner that always executes every suite and then aggregates exit codes (e.g. run each cypress run in a loop or script; collect results; exit non-zero if any suite failed), so that:
  • Every suite runs even if a previous one failed.
  • JUnit/reporting still gets results from all suites.
  • CI gets a single, accurate pass/fail that reflects the full matrix.