We’re currently using Node SDK 4.x and studying whether to move to ottoman.
In our Node SDK based implementation, we manage errors like “cluster closed” or “timeout” by closing the current connection to the cluster and reopening it at next request .
This is to workaround the case where the node to which the application is connected is having some issues.
The sdk should handle all those errors.The application should not need to do anything. ( Except for “cluster closed” - which sounds like the application closed the cluster )
Handle the errors? Yes. The SDKs internally close and reopen connections if the connections are the issue, so it’s not necessary for the application to close and reopen the cluster.
But if the operation cannot be completed within the timeout, the operation is going to return a timeout - if you want the SDK to retry longer, then just increase the timeout.