Improve logging in case of call failure

Previously, we were logging the retry attempt before we even evaluated
whether or not we are going to retry again. Instead, we now compute
the number of failed attempts and either log and bail from the function
or inform the user that we are going to retry the call.

We also include the name of the failed call and the reason why it failed
in case we are retrying it. Once we have exhausted the the number of
retries, we return all errors anyway so there is no reason to log them.
It is up to the caller to perform error handling.

Author: thomaseizinger

src/client.rs

@@ -52,12 +52,17 @@ macro_rules! impl_inner_call {
                     return res;
                 },
                 Err(e) => {
-                    warn!("call retry:{}/{} {:?}", errors.len() + 1 , $self.config.retry(), e);
-                    errors.push(e);
-                    if errors.len() as u8 > $self.config.retry() {
+                    let failed_attempts = (errors.len() + 1) as u8;
+
+                    if failed_attempts > $self.config.retry() {
+                        warn!("call '{}' failed after {} attempts", stringify!($name), failed_attempts);
                         return Err(Error::AllAttemptsErrored(errors));
                     }
 
+                    warn!("call '{}' failed with {}, retry: {}/{}", stringify!($name), e, failed_attempts, $self.config.retry());
+
+                    errors.push(e);
+
                     // Only one thread will try to recreate the client getting the write lock,
                     // other eventual threads will get Err and will block at the beginning of
                     // previous loop when trying to read()