The scaler is multi-threaded (monitor and demand threads); invoke() forked
and, in the child, logged, snprintf'd and setenv()'d before execl(), which
can deadlock on a lock another thread held at fork time and block the
caller in waitpid() forever.
run_script() now posix_spawn()s `sh <script> -f <in.json>`: stdout and
stderr go to the output file through file actions, the provider variables
(PRO_CONF_DIR, PRO_DATA_DIR, PRO_LOG_DIR, PROVIDER_NAME) are passed as the
child's envp (provider_env()), the child starts with an empty signal mask
and default dispositions, and waitpid() retries on EINTR.
The scaler unit suite (scaler/testsuite) was never built; it is now the
meson test "autoscaler test", linked with -Wl,--wrap=fork so it counts
fork() calls during a provider call.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The Nix build previously fetched Go modules from goproxy.cn and Python
build deps from PyPI during the build phase, relying on network leaking
into the sandbox. This is non-reproducible and breaks whenever the
sandbox resolver is unreachable (e.g. pc13v's loopback dnsmasq).
Go:
- Vendor all three modules (frontend/apiserver, frontend/xcecli,
autoscaler/providers/baidu/src/bcctool) and commit vendor/ trees.
- tools/go_build now builds with GOFLAGS=-mod=vendor and GOPROXY=off
instead of GOPROXY=https://goproxy.cn.
Python (xceadmin):
- Add setuptools + wheel to xce.nix nativeBuildInputs.
- meson-build.sh installs with pip --no-build-isolation --no-index
--no-deps so no PyPI access happens at build time.
xceadmin install script:
- Invoke meson-install.sh via find_program('bash') instead of relying
on its #!/usr/bin/env bash shebang, which fails in the Nix install
sandbox (no /usr/bin/env -> 'could not be run').
Cleanup:
- Drop now-unneeded cacert and impureEnvVars=[HOME] from xce.nix and
xcecli.nix (they only existed to enable the impure network fetch).
Docs:
- README build guide: drop '--option sandbox false' from nix build /
nix bundle commands, note the build is now hermetic, and document the
go mod vendor + git add vendor/ maintenance workflow.
Verified: nix build .#xcecore, .#xcecore_debug, .#xcecli, and .#xce all
succeed with the sandbox enabled.
