Description
We have cases where an error can be overcomed and the connector able to proceed further. Such cases should be reported for a user, because they can appear due to a configuration error. Also now we have the background job: periodical cluster discovery for cluster client, which discards incorrect results. A user may expect that a next scheduled discovery was proceeded successfully if there are no warnings and exclude old nodes from a cluster.
We need to support logging by default (are there a mechanism in Java for that?) and provide the ability to change logger: to filter, silent, redirect or change a format of logs.
Also I propose to add logging for discarded cluster discovery results in the scope of this feature. Don't sure about retried transient errors in cluster client: maybe it worth to warn about them too.