Issue Leadership for quorum queues unexpectedly migrates to a single broker node in a cluster, even when nodes are healthy and no network partitions exist. This concentration of leadership leads to high publish latency and consumer lag on the overloaded node.
Symptoms
Nodes communicate in a cluster over the Erlang Distribution Port. If the volume of Raft heartbeats, management traffic, or message synchronization exceeds the RABBITMQ_DISTRIBUTION_BUFFER_SIZE, the port is marked "busy." This delays critical heartbeats, triggering Raft election timeouts and causing leadership to shift to the first available node.
RABBITMQ_DISTRIBUTION_BUFFER_SIZE environment variable. A recommended value for high-traffic environments is 256000 (256MB).#### with the target vhost name.See the RabbitMQ Runtime Documentation for further tuning options.
To speak with a customer representative or a Support Engineer see . Scroll to the bottom of the page and click on your respective region.