Skip to main content

A Postgres SQL Formatter

Project description

CLI application for rebalancing queues in a RabbitMQ cluster

Package Version Build Status Code Coverage BSD


pip install rmq-cluster-rebalance


When you specify the queue master location in RabbitMQ with value of min-masters, RabbitMQ will attempt to balance the distribution of the master locations for queues across the cluster. However, when you restart a RabbitMQ node in a cluster with HA queues, the master location for a queue on that node will move to a different node. Given enough such events and depending on your HA policies, you can end up with an uneven distribution of queues with a master location on a minority of nodes. In this scenario, the nodes in cluster may not be utilized as evenly, with higher CPU, memory, disk, and network utilization on nodes that carry the larger quantity of master nodes.

Rebalancing queues across nodes can also be useful when adding nodes to a cluster with no queue_master_locator configuration set (which uses the default value of client-local), regardless of whether you use HA queues or not.


The application use the RabbitMQ management UI to iterate through each queue in the cluster. The queue is inspected and a policy named rmq-cluster-rebalance is created using the existing policy configuration used by the queue with the addition or replacement of ha-mode, setting it to all and removing ha-params and queue-master-locator if they are set. Once the queue has fully replicated across all nodes in the cluster, the rmq-cluster-rebalance policy is replaced with a new policy that setting ha-mode to nodes and specifying only the new master node in ha-params. When the queue has fully moved, the policy will then removed and the application will move on to the next queue.

Nodes are assigned in a simple round-robin ordering. If a queue is already living on the node where it would be assigned to, it will be skipped and no work is performed on that queue.


This approach should be safe for production use without interruption of publishers or consumers. While it has worked for me without issue, your mileage may vary.

This application requires a relatively current version of RabbitMQ and was only tested with 3.7.16.

CLI Usage

usage: rmq-cluster-rebalance [-h] [-u USERNAME] [-p PASSWORD] [--vhost VHOST]
                             [-L LOG_FILE] [-v] [--debug] [--version]

Rebalances the queues in a RabbitMQ cluster

positional arguments:
  URL                   The RabbitMQ Management API base URL (default:

optional arguments:
  -h, --help            show this help message and exit
  -u USERNAME, --username USERNAME
                        The RabbitMQ Management API username (default: guest)
  -p PASSWORD, --password PASSWORD
                        The RabbitMQ Management API password (default: guest)
  --vhost VHOST         The RabbitMQ VHost to use (default: /)
  --version             output version information, then exit

Logging options:
  -L LOG_FILE, --log-file LOG_FILE
                        Log to the specified filename (default: STDOUT)
  -v, --verbose         Increase output verbosity (default: False)
  --debug               Extra verbose debug logging (default: False)

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

rmq-cluster-rebalance-1.0.0.tar.gz (7.6 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page