After a connection failure, like server restart, the client need to reconnect with the next invocation attempt.
During this phase it happen that the first thread will start the reconnect and wait for the result. The second thread detect that the reconnects are started and skip this.
Unfortunately this cause errors as the thread is not blocked until the first connection is available again.

Actual results:
Failure of inocation with ERROR messages

Expected results:
Wait until connections are available, or fail if reconnect failed