Reconnect an installed Appliance

Sign in as a Management administrator. Find the artifact bundle downloaded after installation and check that it belongs to the intended appliance. It contains sensitive cluster credentials, so retain it in controlled storage outside Management.

Management must be able to reach the appliance nodes. Finish or cancel the current installation and close its draft before reconnecting. Failed installation steps have their own recovery controls on the installation page.

If Management was lost and replaced, use the Management recovery procedure to reconstruct it. If the appliance itself was lost, use backup recovery.

If a cancelled installation is still retained, uploading a bundle closes it and removes its retained input and logs. Download any evidence you still need before uploading.

The connection indicator in the Management header shows the remaining six-hour access period. Open it to see the expiry time and the Renew access and Evict bundle actions.

Ordinary use and browser sign-in do not extend the lease. With two hours or less remaining, Management warns before starting affected operations. Renew first, or choose Accept and start to accept interruption risk for that operation. Accepting the warning does not extend access. Once access expires, use Installation > Reconnect to upload the bundle again.

Keep a verified artifact bundle outside Management before ending access. Open the header’s cluster-access menu, select Evict bundle, review the cancellation warning and confirm. This removes Management’s retained bundle and attempts to close external Kubernetes and node-registry access. It cancels running GitOps and update jobs. It does not uninstall the appliance, undo completed changes or revoke credentials in downloaded artifacts.

Expiry removes access in the same way. Some already-started operations and backup transfers are not explicitly cancelled, but closing network access can still make them fail. An interrupted update requires a new valid bundle and an explicit Retry; access renewal does not resume it automatically. Full installation and appliance recovery have their own attempt lifecycle.

Management distinguishes local removal from remote firewall closure. If it reports Access removed from Management; firewall closure unconfirmed, it could not confirm closure on every node. Reconnect with the saved bundle to reconcile access. Normal shutdown attempts cleanup, but a crash, host loss or unreachable node can prevent it. Do not treat the countdown as a guarantee of remote closure when Management is unavailable.

If the bundle could not be accepted, check both the selected file and node reachability. Select the artifact archive downloaded at installation completion; a release bundle or appliance backup cannot substitute for it. Renewal requires the currently connected appliance’s bundle. End the existing connection before connecting another appliance.

Management can establish access with one reachable authenticated node; it does not need every node to be reachable just to connect. Operations can still need additional nodes, services or quorum. If the bundle is valid but the cluster cannot be verified, check the reported endpoint, network path and node readiness. Then select Try reaching the cluster again. An artifact bundle provides credentials; it does not repair an unreachable cluster.

If Management reports low storage or cannot determine capacity, resolve the storage problem before uploading. For persistent state verification errors, collect diagnostics.