SimpleAsyncHTTPClient's timeout errors are not very informative: we distinguish connect timeouts from other timeouts, but nothing else. It would be helpful to provide more detail, in particular to distinguish DNS resolution from connecting to the HTTP host (which are both part of the "connect" phase).
A motivating example is #3520 in which a slow DNS configuration in kubernetes was responsible for connect timeouts, but this fact was not apparent.