Before node upgrades, the most easily overlooked thing isn’t whether there’s a new version—it’s which role you’re actually running. The documentation places Provisioner, full, and archive on the same network, but their features are different. Confirm the role first, then discuss the upgrade steps; this prevents using one set of instructions for a certain node type to cover another type of node.
The second table should record “what will be kept.” Upgrade materials usually indicate that keys, the environment, and the chain state are typically retained. That means the upgrade plan can’t treat them as ordinary temporary files, nor can it focus only on whether installation succeeds. The third table should record “what may be rebuilt”: configuration, service unit files, and related content may be regenerated by the installer. While they’re not as core as the chain state, they directly affect how the node starts and which runtime parameters it uses.
So a verifiable upgrade checklist should include at least three parts: first, verify the role and features; second, back up or confirm the handover of keys, the environment, and chain state; and last, review the configuration and service units generated by the installer. If you only see “upgrade complete” without confirming these boundaries item by item, the conclusion is still incomplete. What’s discussed here is maintenance planning—not a guarantee of the outcome of any one upgrade.
This applies to maintaining and operating node @Dusk , and the same approach applies to judging the network status of $DUSK : the status should be tied to specific roles and supporting evidence. #dusk can be used to track progress, but it can’t replace the checks done before the upgrade.
Before upgrading, you should also translate “roles differ” into the operational checklist, not leave it at the conceptual level. Provisioner, full, and archive use the same network, but that doesn’t mean their features, configurations, and service units are interchangeable. After every change, separately verify whether the chain state is carried over, whether configuration was rebuilt, and whether services start as expected—only then will the conclusion have a sound basis.
The second table should record “what will be kept.” Upgrade materials usually indicate that keys, the environment, and the chain state are typically retained. That means the upgrade plan can’t treat them as ordinary temporary files, nor can it focus only on whether installation succeeds. The third table should record “what may be rebuilt”: configuration, service unit files, and related content may be regenerated by the installer. While they’re not as core as the chain state, they directly affect how the node starts and which runtime parameters it uses.
So a verifiable upgrade checklist should include at least three parts: first, verify the role and features; second, back up or confirm the handover of keys, the environment, and chain state; and last, review the configuration and service units generated by the installer. If you only see “upgrade complete” without confirming these boundaries item by item, the conclusion is still incomplete. What’s discussed here is maintenance planning—not a guarantee of the outcome of any one upgrade.
This applies to maintaining and operating node @Dusk , and the same approach applies to judging the network status of $DUSK : the status should be tied to specific roles and supporting evidence. #dusk can be used to track progress, but it can’t replace the checks done before the upgrade.
Before upgrading, you should also translate “roles differ” into the operational checklist, not leave it at the conceptual level. Provisioner, full, and archive use the same network, but that doesn’t mean their features, configurations, and service units are interchangeable. After every change, separately verify whether the chain state is carried over, whether configuration was rebuilt, and whether services start as expected—only then will the conclusion have a sound basis.