Description
> Does that mean you have seen it happening on your configurations?
No I just meant the previous issues here.
Lets check also if other sdk clients use the keep-alive option or not.
I'm not aware of any low level TCP settings in any of the other SDKs.
Could you please also help in fixing the failing ci / lint (pull_request) check? How can this be fixed?
I don't think we'll be merging this PR in its current form. I think exposing socket_options is a better option here and I can make a PR next week or so with tests etc.
Tangentially, I'm still not convinced tweaking TCP settings is the right thing to do here. Network failures can happen due to whatever reason and that's not the abstraction layer the SDK should care about. What it should do is handle network failures gracefully/retry if possible etc. I will think more about this and see what to do here.
Originally posted by @sl0thentr0py in #1323 (comment)