Open
Description
Hi,
I'm currently testing arangodb v3.12.5 on Wolfi, and I'm seeing a crash when trying to start the arangod service.
Here's how I'm building the package:
- name: Configure the build
runs: |
cmake \
-DSTATIC_EXECUTABLES=OFF \
-DOpenMP_C_FLAGS="-fopenmp" \
-DOpenMP_CXX_FLAGS="-fopenmp" \
-DOpenMP_C_LIB_NAMES="omp" \
-DOpenMP_CXX_LIB_NAMES="omp" \
-DOpenMP_omp_LIBRARY="/usr/lib/libomp.so" \
-DOpenMP_ROOT="/usr/lib" \
-DCMAKE_C_COMPILER=clang-19 \
-DCMAKE_CXX_COMPILER=clang++ \
-DCMAKE_C_COMPILER_AR="/usr/bin/llvm-ar" \
-DCMAKE_CXX_COMPILER_AR="/usr/bin/llvm-ar" \
-DCMAKE_C_COMPILER_RANLIB="/usr/bin/llvm-ranlib" \
-DCMAKE_CXX_COMPILER_RANLIB="/usr/bin/llvm-ranlib" \
-DCMAKE_EXE_LINKER_FLAGS="-Wl,--build-id=sha1 -fno-stack-protector -fuse-ld=lld -ljitterentropy" \
-DCMAKE_SHARED_LINKER_FLAGS="-fuse-ld=lld" \
-DUSE_LIBUNWIND=true -DUSE_PRECOMPILED_V8=false \
-DCMAKE_POSITION_INDEPENDENT_CODE=ON \
-DCMAKE_C_FLAGS="-fPIC" \
-DCMAKE_CXX_FLAGS="-fPIC" \
-DSNAPPY_INCLUDE_DIR=/usr/include \
-DJEMALLOC_INCLUDE=/usr/include \
-DZLIB_INCLUDE_DIR=/usr/include \
-DUNWIND_INCLUDE_DIR=/usr/include \
-DOPENSSL_ROOT_DIR=/usr/lib \
--preset community
- name: Build
runs: |
cmake \
--build \
--preset community \
--target arangodb \
-- -j$(nproc) VERBOSE=1Here's the command I'm using to start it:
sudo -u arangodb arangod -c /etc/arangodb3/arangod.conf --default-language-check=false --server.endpoint tcp://127.0.0.1:8529 '--server.authentication=false' /tmp/database-dir
I'm seeing the following crash:
2025-06-19T18:44:26.615695Z [655-1] INFO [cf3f4] {general} ArangoDB (version 3.12.5 [linux]) is ready for business. Have fun!
/usr/bin/../lib64/gcc/x86_64-pc-linux-gnu/13/../../../gcc/x86_64-pc-linux-gnu/13/include/c++/bits/atomic_base.h:834: __pointer_type std::__atomic_base<arangodb::containers::ThreadOwnedList<arangodb::async_registry::Promise>::Node *>::load(memory_order) c
onst [_ITp = arangodb::containers::ThreadOwnedList<arangodb::async_registry::Promise>::Node *]: Assertion '__b != memory_order_release' failed.
2025-06-19T18:44:27.413154Z [655-25] FATAL [a7902] {crash} 💥 ArangoDB 3.12.5 [linux], build-id e8ddafe21edac8e2495bd2def3ec6555571269ee, thread 25 [arangod] caught unexpected signal 6 (SIGABRT, sub type unknown) from pid 655 in state "in wait" in direct
ory "/home/build": signal handler invoked - image base address: 0x00005623d1d02000 - CPU context: rip: 0x00007f5776ca002c, rsp: 0x00007f56cd7d8ba0, efl: 0x0000000000000246, rbp: 0x00007f56ed2eb010, rsi: 0x0000000000000363, rdi: 0x000000000000028f, rax: 0
x0000000000000000, rbx: 0x0000000000000363, rcx: 0x00007f5776ca002c, rdx: 0x0000000000000006, r8: 0x0000000000000000, r9: 0x0000000000000000, r10: 0x0000000000000000, r11: 0x0000000000000246, r12: 0x00007f56ed2eb040, r13: 0x00007f56ed2eb000, r14: 0x00000
00000000006, r15: 0x0000000000000000
2025-06-19T18:44:27.413191Z [655-25] INFO [c962b] {crash} Backtrace of thread 25 [arangod]
2025-06-19T18:44:27.414836Z [655-25] INFO [308c3] {crash} frame 1 [+0x0000000006c1635f] (anonymous namespace)::crashHandlerSignalHandler(int, siginfo_t*, void*) [clone .llvm.13269059078790395198] (+0x6f)
2025-06-19T18:44:27.415126Z [655-25] INFO [308c3] {crash} frame 2 [+0x00002933a4f48ae0] __sigaction (+0xe0)
2025-06-19T18:44:27.415331Z [655-25] INFO [308c3] {crash} frame 3 [+0x00002933a4f9e02c] __pthread_kill_implementation (+0x10c)
2025-06-19T18:44:27.415506Z [655-25] INFO [308c3] {crash} frame 4 [+0x00002933a4f489a6] gsignal (+0x16)
2025-06-19T18:44:27.415729Z [655-25] INFO [308c3] {crash} frame 5 [+0x00002933a4f31345] abort (+0x26)
2025-06-19T18:44:27.415963Z [655-25] INFO [308c3] {crash} frame 6 [+0x00002933a51ab26c] std::__glibcxx_assert_fail(char const*, int, char const*, char const*) (+0x61)
2025-06-19T18:44:27.417053Z [655-25] INFO [308c3] {crash} frame 7 [+0x00000000042a68ec] std::thread::_State_impl<std::thread::_Invoker<std::tuple<arangodb::async_registry::Feature::PromiseCleanupThread::PromiseCleanupThread(unsigned long)::{lambda(std::
stop_token)#1}, std::stop_token> > >::_M_run() (+0x3fc)
2025-06-19T18:44:27.417442Z [655-25] INFO [308c3] {crash} frame 8 [+0x00002933a51eeb94] execute_native_thread_routine (+0x14)
2025-06-19T18:44:27.417632Z [655-25] INFO [308c3] {crash} frame 9 [+0x00002933a4f9c3ed] start_thread (+0x2ad)
2025-06-19T18:44:27.417780Z [655-25] INFO [308c3] {crash} frame 10 [+0x00002933a5019dfc] __clone3 (+0x2c)
2025-06-19T18:44:27.417859Z [655-25] INFO [ded81] {crash} available physical memory: 53022552064, rss usage: 352305152, vsz usage: 4173643776, threads: 120
Aborted
This wasn't happening with v3.12.4.4.
I'm happy to provide more details if needed. I'm also starting to debug this further on my side, so I'll keep this issue updated with my findings.
Thanks.
Metadata
Metadata
Assignees
Labels
No labels