Enough stable system handles ~5 thousand WS requests per day and the 1%(~50 connections) connections are closing on an error with the next exception: during the resume proce
