Good Catch! Just tried the sleep() test with add_numbers and the add_numbers message is now showing up in LogzIO. 
I see that Persistent servers is a Business feature, which is a little beyond my budget right now.
I’ll actually poke around the internet. I am hoping that since I am using the python logging framework, with a custom handler provided by LogzIO, there are similar examples of asynchronous logging with things like AWS Lambda Functions or Open Whisk… where the server is short-lived, but similar pattern where messages are queued up and need to be sent before process dies. I’m hopeful I can find a hook to inject some custom code force the LogzIO handler to flush.
You definitely helped confirm the issue, so thanks!
Cheers,
Tyler