Description
currently relp connections use a persistent connection that is once established and will only be established on an error. this is good performance wise for lsh_01 but for balancing queues relp server that receives data this is not sufficient.
Use case or motivation behind the feature request
in case multiple lsh_01 instances end up into a single relp server by a network load balancer bad decisions, this will cause the said relp server to receive all the traffic and rest of the load balancer target servers none and therefore the performance will be degraded.
Description currently relp connections use a persistent connection that is once established and will only be established on an error. this is good performance wise for lsh_01 but for balancing queues relp server that receives data this is not sufficient.
Use case or motivation behind the feature request in case multiple lsh_01 instances end up into a single relp server by a network load balancer bad decisions, this will cause the said relp server to receive all the traffic and rest of the load balancer target servers none and therefore the performance will be degraded.
Related issues
Additional context