gocql fails to refresh node IPs after the Cassandra cluster/node restart

[andrey-dubnik]

Expected Behavior

After the Cassandra cluster maintenance all/some nodes would get new IP addresses the expectation is for the client to re-connect seamlessly.

Actual Behavior

Due to a bug in the gocql new IP addresses of Cassandra are not getting refreshed by the gocql and after the cluster maintenance each temporal service with a dependency on gocql have to be restarted explicitly.

Please consider updating to the latest gocql (1.4.0+) which does have the issue resolved (hopefully) as per the cross-reference of the mentioned issue with gocql PRs.

Steps to Reproduce the Problem

Specifications

• Version: (go.mod have v1.3.1)
• Platform: x86

[mindaugasrukas]

Thanks for reporting! Looking into that.