Switch to blocking mutex lock

[apezel]

Switch to a blocking mutex lock to prevent "Unable to ge transport" error in multi threaded applications.
references issue #230

[AlexPikalov]

This solution seems to be reasonable but I'd like to investigate how it behaves during executing a large number of parallel request. Just don't want to bump into a panic.

Could you please provide some info about how many simultaneous threads did you have and how many nodes?

BTW Some time ago I was thinking about providing a very specific load balancer strategy which would always return the same single transport (node). I think a cluster consisting of a single node might be a quite common case so any calculations might be omitted for such a simple case.

If it's sort of your case and/or you are interested in that feel free to create an issue for implementation of such strategy and eventually submit a PR.

[AlexPikalov]

Please give me an another day to play around this solution and we'll merge it.

Thanks for your PR

[apezel]

Do you mean something like this ?

[AlexPikalov]

Exactly 👍 Thank you

As for the solution with lock instead of try_lock it's okay since lock panics if

This function might panic when called if the lock is already held by the current thread.

which seems to be non-achievable because method next contains only one lock and a lock is released immediately after next execution finishes

[AlexPikalov]

Thank you for your PR