Updated on 2025-09-03 GMT+08:00

Does Specification Modification Affect Services?

Yes. See Table 1.

Table 1 Impact of specification modification

Modified Object

Impact

Brokers

  • During the broker increase, the RabbitMQ workload balance process is restarted, triggering a master/standby switchover and causing services to temporarily stutter. Ensure that your client can auto-reconnect. Modify specifications during off-peak hours.
  • The change duration depends on the number of added brokers, and is 5–10 minutes per broker. The total change duration = Single change duration × Number of added brokers.

Storage space

  • Storage space expansion does not affect services.
  • You can expand the storage space 20 times for each instance.
  • Available storage space = Actual storage space – Storage space for storing logs – Disk formatting loss For example, if the storage space is expanded to 700 GB, the storage space for storing logs is 100 GB, and the disk formatting loss is 7 GB, then the available storage space after capacity expansion will be 593 GB.
  • The total change duration is within 5 minutes.

Broker flavor

  • For single-node RabbitMQ 3.x.x instances, nodes are restarted and services may temporarily stutter in minutes during the modification. Ensure that your client can auto-reconnect. Modify specifications during off-peak hours.
  • For cluster RabbitMQ 3.x.x instances not configured with mirrored/quorum queues, nodes are restarted in sequence and services may stutter in minutes during the modification. Ensure that your client can auto-reconnect. Modify specifications during off-peak hours.
  • For cluster RabbitMQ 3.x.x instances not configured with mirrored/quorum queues, nodes are restarted in sequence and services may stutter for several seconds during the modification. Ensure that your client can auto-reconnect. Modify specifications during off-peak hours.
  • During the modification of cluster RabbitMQ 3.x.x instances, connections to the changed nodes will be switched to other nodes, posing overload risks (for example, connections excess or memory high watermark) on them. You are advised to use them within the instance specifications. For more information, see Specifications.
  • Persistent exchanges, queues, and messages are required. Otherwise, messages may be lost after node restarts. For details, see Configuring RabbitMQ Persistence.
  • Brokers will be restarted in sequence during an instance change, which depends on the number of brokers. The change duration per single broker is 5–10 minutes. The total change duration = Single change duration × Number of brokers.

The following example shows how to configure message retry on a Java client.

ConnectionFactory connectionFactory = new ConnectionFactory();
//Set a service address.
connectionFactory.setHost("localhost");
//Set a port.
connectionFactory.setPort(5672);
//Auto retry:
connectionFactory.setAutomaticRecoveryEnabled(true);
connectionFactory.setNetworkRecoveryInterval(5);
connectionFactory.setTopologyRecoveryEnabled(true);