Confd behind cluster ip connection issues

I’ve confd running in a single node kubernates cluster. And I have multiple client connecting to confd via a service IP. Things work fine initially. But after about 40 minutes, the connection between confd and service IP disappears once confd starts sending some data on subscription socket. Connection between client and service IP remains intact, with client waiting forever on subscription socket.

Is there some cdb API which would help us in detecting this case from client side and help in reconnecting?


No API for detecting such an issue where the client and ConfD think they are still connected. It would need to be handled at the “service IP” level to close the connection to the client if the connection to “disappears”.