Should mk_event_wait handles EINTR?

[lobshunter]

Bug Report

What did you do?
I ran fluent-bit kubernetes daemonset with HTTP server on, it starts a monkey server inside.

What did you see?
HTTP server of fluent-bit exits several seconds after fluent-bit started, but fluent's log scraping still works.
However, it leads to kubernetes liveness probe failure, hence fluent-bit will be SIGTERM killed again and again.

What do you know about this?
After a lot of digging I found that epoll_wait here does not handles negative return value, so mk_event_wait here returns unexpectedly, then HTTP server exits too.

In my case, the epoll_wait returns -1 and errno is EINTR. According to epoll_wait(2)

The call was interrupted by a signal handler before either
(1) any of the requested events occurred or (2) the timeout expired.

I found a solution in stackoverflow which ignores the EINTR error. That works for me.

So I wonder if this is a bug of monkey? And maybe other mk_event_wait implementations like select should handle such return value too.