The error I am seeing is:
TASK [rabbitmq : Enable feature flags] *****************************************
Wednesday 20 March 2024 13:33:14 +0000 (0:00:00.099) 0:00:21.630 *******
ok: [controller0] => (item=quorum_queue)
ok: [controller0] => (item=implicit_default_bindings)
ok: [controller0] => (item=virtual_host_metadata)
ok: [controller0] => (item=maintenance_mode_status)
ok: [controller0] => (item=user_limits)
ok: [controller0] => (item=stream_queue)
ok: [controller0] => (item=classic_queue_type_delivery_support)
ok: [controller0] => (item=tracking_records_in_ets)
ok: [controller0] => (item=stream_single_active_consumer)
ok: [controller0] => (item=listener_records_in_ets)
ok: [controller0] => (item=feature_flags_v2)
ok: [controller0] => (item=direct_exchange_routing_v2)
ok: [controller0] => (item=classic_mirrored_queue_version)
ok: [controller0] => (item=stream_sac_coordinator_unblock_group)
failed: [controller0] (item=restart_streams) => {"action": "community.rabbitmq.rabbitmq_feature_flag", "ansible_loop_var": "item", "changed": false, "item": "restart_streams", "msg": "restart_streams feature flag is not available"}
This is when attempting to upgrade rabbitmq. We seem to have 3.11.28 in 2023.1, but restart_streams wasn't added until 3.12:
https://www.rabbitmq.com/docs/feature-flags#core-feature-flags
It seems like in the upgrade job we do a reconfigure prior to the upgrade (which essentially gives us the master containers). This is not a documented step.
The reconfigure step was added when we were using HA queues. Given we are now instead enabling quorum queues by default, we do not need this reconfigure step anymore. See docs here for the required steps: https:/