Honestly, I don't think it's anything outside the usual suspects, but hopefully this brain dump is useful.
1) compiling is slowwwww, and something about my workplace environment (docker builds) means there's a lot of downloading and a lot of disk space used. cargo clean is helpful, and I blame docker more than Rust here. I am running in a VM, but I've had to increase the allocation to 150 G to run a fairly barebones Linux environment (no desktop) and not have to clean up my environment frequently. Some of my other environmental stuff is also super slow, so I'm pretty grumpy about speed in general.
1a) I've had cargo fmt eat my files when I was out of disk space. It created a new file, but the file length was 0, and my old file was gone. (I did not open a bug report)
1b) I hate how my dependency tree has about 700 items, but I still have to update h2 for a http/2 server security bug to make cargo audit happy, when AFAIK, we don't even use http/2 at all. (we do use hyper, and aws-sdk)
1c) There's about 7 different ways to get at epoll from Rust, but I don't think any of them let you set a sub-ms timeout, unless you use a timerfd (at least one of the 7 ways does do the timerfd for you, but that doesn't make it better, just easier). Why doesn't anyone make epoll_pwait2 available somehow. Although, really, this is more of a question of why didn't Linux offer a useful timeout in epoll_wait to begin with. If I were more familiar with Rust, maybe I'd be confident in just doing it myself, like I can in other languages, but it still seems scary.
2) It's pretty difficult to figure out what my system is doing from the outside. But I'm used to Erlang's introspection, and I was tempted to think that some of the features of Rust would make it Erlang-ish. Maybe I just need to use tokio_console, but it's not there by default, and I'm not sure how I would connect up to my running production system. This is probably more of a me problem or a my environment problem than a Rust problem.