RabbitMQ 3.8: basic.consume fails early on node boot before QQ leader is elected

[techblack]

Describe the bug

Using single-node rabbitmq's Quorum Queues (qq), but leader not known appears. Command will be forwarded once leader is known.

• Rabbitmq version : 3.8.34
• one vhost
  

Reproduction steps

1. docker run rabbitmq:3.8.34-management-alpine
2. delcare a vhost and qq
3. restart
   ...

Expected behavior

qq always in leader

Additional context

using rabbitmqctl eval "ra:trigger_election('scanner_00000000000000000000000000000001')." can fix temporary it .
How should I completely resolve this issue?

[techblack]

This problem is not inevitable, but it is easy to occur on some machines.

[michaelklishin]

RabbitMQ 3.8 is completely out of support. Upgrading is the only thing we can recommend.

[michaelklishin]

There is a hint in the stack trace. This happens in a basic.consume method handler, so delaying registering consumers on newly created QQs may serve as a temporary solution until you can upgrade to 3.12.10.