ci(test): distinguish "skipped due to missing tools" from "tested and passed" in startup script tests

[bitflicker64]

Problem

The three startup regression test scripts exit 0 when required tools are missing from the environment:

# test-start-hugegraph.sh, test-start-hugegraph-pd.sh, test-start-hugegraph-store.sh
if ! command -v "$tool" >/dev/null 2>&1; then
    echo "SKIP: required tool '$tool' not found — skipping test suite"
    exit 0
fi

Exit code 0 is indistinguishable from "all tests ran and passed". If a CI runner is missing lsof, curl, java, or crontab, the step silently reports success and no one notices the tests never ran.

This was flagged as a non-blocking follow-up in the review of #3044 and #3047.

Expected behavior

When required tools are missing, the test scripts should exit with a distinct code (e.g. 77, the conventional POSIX skip code used by Automake and TAP) so CI can tell the difference between:

• exit 0 — tested and passed
• exit 77 — skipped due to missing environment
• exit 1 — tested and failed

The CI workflow steps should handle 77 explicitly and emit a ::notice:: annotation instead of failing.

Files to change

Test scripts — change exit 0 to exit 77 in the tool-check block:

• hugegraph-server/hugegraph-dist/src/assembly/travis/test-start-hugegraph.sh (lines 148–153)
• hugegraph-server/hugegraph-dist/src/assembly/travis/test-start-hugegraph-pd.sh (lines 132–137)
• hugegraph-server/hugegraph-dist/src/assembly/travis/test-start-hugegraph-store.sh (lines 113–127)

CI workflow steps — wrap the test call to handle exit 77:

• .github/workflows/server-ci.yml — "Run start-hugegraph.sh foreground mode tests" step (lines 73–79)
• .github/workflows/pd-store-ci.yml — "Run start-hugegraph-pd.sh foreground mode tests" step (lines 110–114) and the equivalent Store step

CI wrapper pattern:

run: |
  set +e
  $TRAVIS_DIR/test-start-hugegraph.sh $SERVER_DIR
  EXIT=$?
  set -e
  if [ $EXIT -eq 77 ]; then
    echo "::notice::Startup tests skipped — required tools not available in this environment"
    exit 0
  fi
  exit $EXIT

Related

• Parent issue: [Bug] Docker containers never restart when Java process crashes — broken cron monitor + tail -f /dev/null supervision #3043
• Flagged in review of fix(server): fix foreground mode exit code propagation in startup scripts #3044 (imbajin comment: "the new startup regression test currently skips with exit 0 when required tools are missing")
• Flagged in review of fix(pd,store): fg mode exit code propagation in startup scripts #3047 (same pattern in PD/Store test scripts)