Yes, this is planned for 3.4.0. It just takes a bit longer as we decided to implement this together with request pipelining ([TINKERPOP-1775](https://issues.apache.org/jira/browse/TINKERPOP-1775)).
However, I'm not sure if it resolves the issue you're describing. Your problem sounds like a connection was closed because it was idle for some time and then the driver tried to use it again. These changes should reduce the number of connections that are opened and especially the number of connections that are idle for a long time, but there will still be situations where connections are idle. I assumed that pings should prevent this from happening. Do you know whether CosmosDB supports pings for its Gremlin API? Could you also please provide a full stack trace? It would especially be interesting to see where in Gremlin.Net the exception originated. [ Full content available at: https://github.com/apache/tinkerpop/pull/903 ] This message was relayed via gitbox.apache.org for dev@tinkerpop.apache.org