Getting java.nio.channels.ClosedChannelException after an idle period of time

[benoit-david]

Expected Behavior

• Connections should be recycled after a period of inactivity.

Current Behavior

• After a long period of inactivity (a night for example) the first Query made to the Gremlin cosmosDB database ends up with java.nio.channels.ClosedChannelException.

After several refresh, the connections are recycled and everything is back on track.

Possible Solution

• Is the KeepAliveInterval feature implemented ? As I don't see it in the GremlinProperties options. The connections should be recycled automatically by the pool.

Steps to Reproduce (for bugs)

• Open a cluster with 2 connections
• Let the cluster idle for a long period (a night for example)
• Try to query the Gremlin CosmosDB database

Snapshot Code for Reproduce

@Repository
public interface MyExampleRepository extends GremlinRepository<FooBar, String> {

}

Then calling a simple method like following would cause the issue after an idle period:

    Optional<FooBar> fooBar = myExampleRepo.findById(myId);
}

Branch

• 2.0.2

Your Environment

• Spring data gremlin version used: 2.0.2
• Gremlin driver version: 3.2.4
• SDK version: 1.8

[superrdean]

Hi @benoit-david
This is a new feature indeed, and in a short time, we may not consider adding it to our library. But PR is welcome. Sorry for the inconvenience.

[chenrujun]

Closing this issue.
Because it's not active for a long time.
If anyone have similar issue, please create issue in new repo: https://github.com/Azure/azure-sdk-for-java/issues