bitflicker64 commented on code in PR #3336:
URL: https://github.com/apache/hugegraph/pull/3336#discussion_r4265614454
##########
hugegraph-pd/hg-pd-dist/src/assembly/static/bin/start-hugegraph-pd.sh:
##########
@@ -60,6 +60,12 @@ LOGS="$TOP/logs"
OUTPUT=${LOGS}/hugegraph-pd-stdout.log
GITHUB="https://github.com";
PID_FILE="$BIN/pid"
+JVM_MODULE_OPTIONS="$BIN/jvm-module.options"
+
+if [[ ! -f "$JVM_MODULE_OPTIONS" || ! -r "$JVM_MODULE_OPTIONS" ]]; then
Review Comment:
Minor: with this check the PD and Store launchers need
`bin/jvm-module.options`, but `test-launcher-gc.sh` (not in this diff) never
copies the PD or Store file into its fixture. It passes only because the
`server` iteration runs first and leaves Server's `jvm-module.options` in the
shared `$FIXTURE/bin` (line 54), so the PD and Store cases start the real JVM
with Server's options instead of their own. Server's file also opens
`java.lang`, which is why nothing fails today.
Evidence at abda50a, macOS arm64, OpenJDK 17.0.20.1: the script as shipped
prints 21 `PASS` lines. The same script with only the loop order changed to
`for component in pd store server` prints `FAIL: pd default launcher rejected
JVM options` and the same line for the other six PD modes and all seven Store
modes; the missing-file exit here is the only launcher path that fits.
Requested change: in the `pd)` and `store)` branches of
`test-launcher-gc.sh`, copy `$source_bin/jvm-module.options` into
`$FIXTURE/bin/` the way the `server)` branch does, so each component runs with
the file it ships.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
