EventStore client does not re-resolve the URI DNS record when re-connecting if the Create() method was used with a load balancer

[LewisCarey2]

Describe the bug
When using the Create() method for connecting to an Event Store cluster via a load balancer the client does not re-resolve the uri dns record when the connection is broken and then tries to reconnect. This impacted us when AWS were replacing the IP addresses associated with an ELB used by Event Store. Despite them changing 1 record of the DNS record set (of 3 IPs) our clients could not reestablish connectivity using one of the other IPs.

In version 6 it seems that the ability to pass in a custom HttpClient will be added, which will allow us to fix the issue in the underlying client, but I thought I should document it here in case others are having a similar issue. I’d also suggest that the default HttpClient be fixed to properly honour DNS changes.

The issue with singleton HttpClients has been documented well here: http://byterot.blogspot.com/2016/07/singleton-httpclient-dns.html.