I’m upgrading a fairly large consul deployment with basic wan federation across multiple datacenters, from a fairly old version (<= 1.2.4), marching up towards latest, with only a few ACL tokens in use and no ACL replication. I’m currently taking the first step, upgrading to 1.2.4, I have several questions, mostly due to what seems to me to be inconsistent documentation.
- Do I need to upgrade the primary/acl_datacenter first, or last? The specific version documentation for 1.2.4 says I need upgrade the ‘primary’ datacenter last. However, the tutorial for upgrading WAN federated clusters (which is written for upgrading from 1.7.11 to 1.9.2, so is a bit ahead of where I am in the process), says that I should upgrade the ‘primary’ datacenter first. Is this just something that will change I as I upgrade to later versions? I do see that the specific upgrade instructions for 1.6.10 to 1.8.x don’t seem to specify any ordering between DCs for the upgrade.I’m going with the version specific instructions for now but I’m very confused. Is it the case that the order doesn’t actually matter because I do not have ACL replication in use?
- Do I need to ‘leave’ on each of my server nodes before starting up with 1.2.4? Again, the upgrade-federated-environment tutorial linked above also says that nodes should
leavethe cluster before restarting consul. This is the only piece of documentation regarding upgrades I’ve seen that says anything about using that command, and given that the default setting forleave_on_terminateis false for server nodes, I’m very confused.
I’m almost certainly misunderstanding something here and I’d appreciate your assistance!