I have no support/technical question and have seen the support category. (Be aware that direct support questions will be deleted.)
on
Which general topic do you have
I’m running Nextcloud with the Recognize app on a VM (VMware, x86_64, Ubuntu 24.04, glibc). Native TensorFlow mode crashes consistently — the classifier process dies with a bus error (exit 135 / SIGBUS). WASM mode works fine, so the app itself is functional. It’s specifically the native libtensorflow path that fails.
Reproduced by hand:
sudo -u www-data ./bin/node -e “require(‘@tensorflow/tfjs-node’)” → Bus error, exit 135.
strace shows it dies right after mmap’ing libtensorflow.so.2.9.1, on first access to the mapped region:
SIGBUS {si_signo=SIGBUS, si_code=BUS_ADRERR, si_addr=0x…}
Already ruled out (with evidence): CPU has AVX/AVX2/AVX-512 (so not an instruction issue), the .so reads end-to-end clean with dd (not corrupt/truncated), no I/O or MCE errors in dmesg, 13 GiB RAM free, THP off makes no difference, /dev/shm is 7.9 GB.
Environment: Nextcloud 33.0.6 Enterprise, Recognize 11.0.1, native install (not Docker/AIO), node 20.9.0 (bundled), 8 vCPU, 16 GB RAM.
Has anyone hit this in a virtualized environment? How did you solve it? Happy to paste full lscpu / cpuinfo / strace.