In short, the issue is that when we manually failover MemoryDB while having 2 or 3 nodes (1 primary, 1 or 2 replicas, 1 shard), we're unable to reach the MemoryDB cluster via the cluster endpoint unless we restart the ECS task (or reconnect using ConnectionMultiplexer).
We've found that if we do:
foreach (var endpoint in connectionMultiplexer.GetEndPoints(false))
{
Console.WriteLine(endpoint.ToString());
}
We only get back the cluster endpoint. But our understanding is that it should be returning all discovered nodes.
Is there anything we can do to further debug this issue?