[bug](jdbc) test_jdbc_connection rpc should check enable_java_support firstly

[zhangstar333]

What problem does this PR solve?

Problem Summary:
if enable_java_support=false, and the rpc of test_jdbc_connection may cause coredump

*** Query id: 0-0 ***
*** tablet id: 0 ***
*** Aborted at 1776056623 (unix time) try "date -d @1776056623" if you are using GNU date ***
*** Current BE git commitID: 000395ded91 ***
*** SIGSEGV address not mapped to object (@0x10) received by PID 3520046 (TID 3524680 OR 0x10cc164ee640) from PID 16; stack trace: ***
 0# doris::signal::(anonymous namespace)::FailureSignalHandler(int, siginfo_t*, void*) at ../src/common/signal_handler.h:417
 1# PosixSignals::chained_handler(int, siginfo*, void*) in /mnt/disk6/common/jdk-17.0.16/lib/server/libjvm.so
 2# JVM_handle_linux_signal in /mnt/disk6/common/jdk-17.0.16/lib/server/libjvm.so
 3# 0x000014D80A03E6F0 in /lib64/libc.so.6
 4# setTLSExceptionStrings at /home/runner/work/doris-thirdparty/doris-thirdparty/thirdparty/src/doris-thirdparty-hadoop-3.4.2.2-for-doris/hadoop-hdfs-project/hadoop-hdfs-native-client/src/main/native/libhdfs/jni_helper.c:908
 5# printExceptionAndFreeV at /home/runner/work/doris-thirdparty/doris-thirdparty/thirdparty/src/doris-thirdparty-hadoop-3.4.2.2-for-doris/hadoop-hdfs-project/hadoop-hdfs-native-client/src/main/native/libhdfs/exception.c:196
 6# printExceptionAndFree at /home/runner/work/doris-thirdparty/doris-thirdparty/thirdparty/src/doris-thirdparty-hadoop-3.4.2.2-for-doris/hadoop-hdfs-project/hadoop-hdfs-native-client/src/main/native/libhdfs/exception.c:226
 7# getJNIEnv at /home/runner/work/doris-thirdparty/doris-thirdparty/thirdparty/src/doris-thirdparty-hadoop-3.4.2.2-for-doris/hadoop-hdfs-project/hadoop-hdfs-native-client/src/main/native/libhdfs/jni_helper.c:824
 8# doris::Jni::Env::GetJNIEnvSlowPath(JNIEnv_**) at ./be/build_ASAN/../src/util/jni-util.cpp:182
 9# doris::Jni::Env::Get(JNIEnv_**) at ../src/util/jni-util.h:57
10# doris::JniReader::open(doris::RuntimeState*, doris::RuntimeProfile*) at ./be/build_ASAN/../src/format/jni/jni_reader.cpp:97
11# doris::PInternalService::test_jdbc_connection(google::protobuf::RpcController*, doris::PJdbcTestConnectionRequest const*, doris::PJdbcTestConnectionResult*, google::protobuf::Closure*)::$_0::operator()() const at ./be/build_ASAN/../src/service/internal_service.cpp:1084 

Release note

None

Check List (For Author)

• Test

  • Regression test
  • Unit Test
  • Manual test (add detailed scripts or steps below)
  • No need to test or manual test. Explain why:
    • This is a refactor/code format and no logic has been changed.
    • Previous test can cover this change.
    • No code files have been changed.
    • Other reason

• Behavior changed:

  • No.
  • Yes.

• Does this need documentation?

  • No.
  • Yes.

Check List (For Reviewer who merge this PR)

• Confirm the release note
• Confirm test cases
• Confirm document
• Add branch pick label

[zhangstar333]

/review

[zhangstar333]

run buildall

[github-actions]

No blocking issues found in this change.

Critical checkpoints:

• Goal / correctness: The goal is to prevent test_jdbc_connection from entering the JNI path when enable_java_support=false. The new early return in be/src/service/internal_service.cpp does that before the request is queued or JniReader::open() can call Jni::Env::Get(), so it addresses the reported crash. FE already converts non-OK PStatus from this RPC into a user-facing error.
• Scope / minimality: The modification is small, focused, and limited to the affected RPC entry point.
• Concurrency: No new concurrency is introduced. The handler now either fails fast on the RPC thread or follows the existing heavy-work-pool path unchanged. No shared-state or lock behavior changes.
• Lifecycle / static initialization: No lifecycle or static-init risk is added. The guard actually reduces exposure to JVM/TLS initialization paths when Java support is disabled.
• Configuration: No new config is added. The change correctly follows the existing enable_java_support contract and restart requirement.
• Compatibility: No FE/BE protocol, symbol, storage-format, or rolling-upgrade compatibility issue is introduced.
• Parallel code paths: Existing HDFS/JNI entry points already guard on enable_java_support; this brings test_jdbc_connection in line with those paths.
• Conditional check quality: The new condition is justified and straightforward because BE only initializes JNI when Java support is enabled in doris_main.cpp.
• Test coverage: No automated test was added for this crash path. That is the main residual risk, but given the very localized fix and the direct alignment with existing guard patterns, I do not see it as a blocker.
• Test results changed: Not applicable.
• Observability: Existing status propagation is sufficient for this path; no additional logging/metrics appear necessary.
• Transaction / persistence / data-write impact: Not applicable.
• FE-BE variable passing: Not applicable.
• Performance: Slightly improved in the disabled-Java case by failing before queueing, thrift deserialization, and JNI setup.
• Other issues: None found.

Summary opinion: The fix is correct, minimal, and consistent with adjacent JNI/HDFS guard patterns. The only notable gap is missing automated coverage for this RPC failure path.

[hello-stephen]

BE UT Coverage Report

Increment line coverage 0.00% (0/7) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	53.06% (20148/37969)
Line Coverage	36.61% (189502/517687)
Region Coverage	32.83% (146964/447719)
Branch Coverage	33.99% (64391/189423)

[hello-stephen]

BE Regression && UT Coverage Report

Increment line coverage 14.29% (1/7) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	73.63% (27379/37183)
Line Coverage	57.28% (295604/516103)
Region Coverage	54.42% (245918/451851)
Branch Coverage	56.13% (106658/190005)