Upgrading RabbitMQ Cluster

[Vincentvwal]

We are planning an upgrade of our RabbitMQ Cluster (which is long overdue) from 3.8.3 and erlang 22.3.4.1 to 3.8.16 and erlang 1.24.0.

During the update of our acceptance machine (single node) we just ran the update and everything is running, however for production we also have messages which are awaiting delivery which we cannot afford to lose.
So I've been testing in a newly setup cluster with some mock data to see how we exactly need to upgrade our cluster and I want to check if i'm doing this right.

Step 1.
I bring the slave node down using rabbitmqctl stop_app and first run an upgrade to 3.8.8 and erlang 1.23.3.1.
After this, the node starts back up and re-joins the cluster

Step 2.
As queues are now out of sync, i manually sync the queues which are.

Step 3.
I bring the master node down using rabbitmqctl stop_app and upgrade this node as well.
After the update, this node also rejoins the cluster but now as slave as the other node has been promoted.

Step 4.
I re-sync the queues again manually.

After this, i repeat the steps again, this time upgrading to 3.8.16 and Erlang 1.24.0 (and make sure we update more regularly)
Is this a correct way to go about upgrading this?

Small notes:
We are running a cluster with a default ha-all policy
Queues will be set up as Lazy queues before the upgrade (cutting down performance, but making it more predictable)
After the upgrade the nodes are taken down again for a configuration change (increase memory allowed usage)