reaper_status_logs = true causes before_worker_boot restart loop with Puma 7.x

[obromios]

Environment

• puma_worker_killer 1.0.0
• puma 7.1.0
• Ruby 3.4.8
• Rails 8.0.3
• Heroku (Linux), 1 worker (WEB_CONCURRENCY=1)

Steps to reproduce

Set config.reaper_status_logs = true in the PumaWorkerKiller.config block:

before_fork do
  require 'puma_worker_killer'
  PumaWorkerKiller.config do |config|
    config.ram = 512
    config.percent_usage = 0.9
    config.frequency = 30
    config.reaper_status_logs = true
  end
  PumaWorkerKiller.start
end

Expected behaviour

PumaWorkerKiller starts normally and monitors worker memory.

Actual behaviour

Puma logs the following warning on every worker boot and restarts workers
every ~30 seconds (matching config.frequency):

WARNING hook before_worker_boot failed with exception (NoMethodError)
undefined method 'call' for true

This occurs 100+ times per hour in production, causing continuous worker
churn and H12 request timeouts on every in-flight request at the moment
of each restart.

Root cause

reaper_status_logs = true appears to register a before_worker_boot
hook using an API that changed in Puma 7.x, resulting in the hook being
set to the boolean true rather than a callable. When Puma attempts to
invoke the hook it calls true.call, which raises NoMethodError.

Workaround

Remove config.reaper_status_logs = true from the config block.

[obromios]

I note the issue was found by Claude whilst debugging a heroku deployment for me. Claude wrote the issue report.