casperisfine
commented
2 years ago Also there's kind of two schools on how to deal with this.
Some libraries insist that it's the application's job to call the proper callback on their libraries when they fork, so in this case HTTPPersistent#reload. Which I admit is cleaner, but also sometimes is hard, as the relevant object are sometimes buried deep in private variables.
Some other libraries instead add a fork detection mechanism (generally @saved_pid != Process.pid), so that they automatically discard connections after fork.
If you are open for the later, I'd be happy to provide a PR.
Ref: https://github.com/mperham/connection_pool/pull/140
When forking a process, you need to close existing connection to avoid sharing them across processes.
shutdowndoes that, but it also mark the pool as no longer usable.connection_pool2.2.4introduced#reloadthat discard existing connections, but let the pool be used again later. It's a much better fit for anafter_forkcallback.