Linux::Event::Fork examples

This directory contains both usage examples and stress tests.

The stress tests are intentionally small and self-contained so you can run them under strace, perf, valgrind, or with different loop backends.

Stress tests

90_stress_timeout_churn.pl

What it stresses

• fork/exit churn (many short-lived children)
• timer scheduling + cancellation (timeout)
• drain-first teardown stability under load

Controls

• N (default 200): number of children
• TIMEOUT (default 0.02): timeout seconds

Expected

• The loop stops cleanly after all children complete.
• Some jobs are marked timedout (timeout callback fired).
• Because the child installs a TERM handler that exits 0, a timed-out job can still count as exit_ok.

Run:

perl -Ilib examples/90_stress_timeout_churn.pl
N=1000 TIMEOUT=0.01 perl -Ilib examples/90_stress_timeout_churn.pl

91_stress_stdin_with_timeout.pl

What it stresses

• backpressure-aware stdin streaming (stdin_write / write watcher)
• timeout firing while stdin is still being written
• teardown stability after EPIPE (child exit closes stdin pipe)

Controls

• TIMEOUT (default 0.05): timeout seconds
• MB (default 5): payload size in MiB written to stdin

Expected

• Prints START, then [timeout], then DONE summary.
• Does not hang, and does not die from SIGPIPE.

Run:

perl -Ilib examples/91_stress_stdin_with_timeout.pl
TIMEOUT=0.03 perl -Ilib examples/91_stress_stdin_with_timeout.pl
MB=20 TIMEOUT=0.05 perl -Ilib examples/91_stress_stdin_with_timeout.pl