HelixObs / client-python / 28056078235

95%

push

github

fix: register at-fork handlers to prevent gRPC deadlock in child processes

Pipelines that use os.fork() (directly or via multiprocessing/cadcutils)
can deadlock permanently when fork() happens while a BatchSpanProcessor or
BatchLogRecordProcessor background thread holds an internal gRPC mutex.
The child inherits the locked mutex but the thread is dead, so any
subsequent gRPC call in the child waits forever.

Add helixobs/_fork.py with os.register_at_fork() handlers:
  before:         force_flush(timeout_millis=500) on all providers — drains
                  the export queue and releases locks before fork copies state
  after_in_child: shutdown() on all providers — stops background threads
                  in the child so it never tries to export telemetry

Both configure_logging(otlp=True) and Instrument.__init__ register their
providers via _fork.register_provider_for_fork(). No-op on Windows where
os.register_at_fork is not available.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

20 of 29 new or added lines in 3 files covered. (68.97%)

385 of 405 relevant lines covered (95.06%)

0.95 hits per line