Updating the EasyUC Manager

What it is for

The EasyUC Manager is a product of its own with its own version number. It consists of the Manager interface, the Manager API with the built-in operator and the definition of the EasyUC resource. An EasyUC update does not change the Manager; it is updated separately.

EasyUC 4.8.7 requires Manager 1.15.0 or later, because only this level knows the EasyUC-Update page and the single image. Therefore update the Manager first and EasyUC afterwards.

There are three paths: the offline package for the Manager, which is applied on the console of the first node; the Operator-Update in the Manager interface; and the signed offline package on the System-Updates page for clusters with exactly three nodes.

Prerequisites

The Manager package of your release from EasyUC. The accompanying documents of the release state which path applies to your package.
For the offline package: console access with administrator rights on the first node.
For the operator update: a login to the Manager. The images of the new Manager must already be on the nodes.
For the signed offline package: exactly three nodes in state Ready, synchronised time on all nodes and VM snapshots of all nodes.
In clusters with several nodes, the Manager image must be present on all nodes so that the Manager can start on another node if a node fails.

Step by step: offline package on the console

Copy the Manager package to the first node.
Unpack it and start the included script apply-manager-update.sh with administrator rights, as described in the accompanying documents.
The script imports the images, distributes the Manager image to the other nodes, updates the definition of the EasyUC resource and runs the Manager update.
Wait for the final check of the script. If the update fails, the script reverts the change itself.
Log in to the Manager again. The version is shown in the navigation below the logo ("Version …").

Step by step: operator update in the interface

Open the user menu at the top right of the Manager and select Operator-Update.
In the Operator Self-Update dialog, select the chart file of the Manager (.tgz or .tar.gz) under Select Helm Chart.
Select Start Upgrade.
Follow the states "Uploading Chart...", "Upgrading Operator..." and "Pod Restarting..." and the Progress Log.
Wait for "Upgrade Completed". The page reloads afterwards.

Step by step: signed offline package (three nodes)

Open the System-Updates page in the Manager. The card EasyUC-Update (signiertes Offline-Paket) is at the top.
Drag the .run file and the matching .run.sig together into the drop area.
Select Hochladen und prüfen and confirm with the Manager password. The upload can be resumed.
Wait for "Signatur gültig". If "Signatur ungültig – Paket nicht verwenden" appears, stop.
Select Auf allen Knoten vorladen. The nodes are prepared one after the other; running services do not change.
In the node table, check the columns Ready, Datei, Vorladen, Vorbereitet, Zeit synchron, Frei and Vertrauensanker.
In the Einspielen section, select Prüfen (--ha-check).
Create VM snapshots of all nodes and tick VM-Snapshots aller Knoten erstellt in Gefahrenbereich: Update einspielen.
Type the confirmation word shown and select Update einspielen.
Leave the page open. The Manager restarts during the run; the page keeps polling the state and then shows result and log.
Finally select Hochgeladene Dateien löschen.

Fields and options

Fields and options
Field / action	Meaning and use
Manager-Version	Running version of the Manager. The addition "von der Vorabprüfung nicht unterstützt" means that the package does not know this level as a starting version.
Koordinator	The node the Manager runs on. The package is uploaded there and distributed from there.
HA-Cluster	The steps of the signed offline package require exactly three nodes.
Vorladen	Imports the content of the package on a node without changing running services.
Zeit synchron	Shows whether the clock of the node is synchronised by NTP. Without synchronisation the update cannot be applied.
Vertrauensanker	Shows whether the public key for signature verification is installed on the node.
Anhalten / Erneut versuchen	Pauses or repeats distribution and preloading.
Cluster prüfen (--ha-verify)	Checks the cluster again after the update has been applied.
Log	Output of the selected step, refreshed every three seconds during a run.

Check the result and troubleshoot

The navigation of the Manager shows the new version below the logo, and all Manager pages are reachable.
The EasyUC-Update page exists. If it is missing, a Manager older than 1.15.0 is still running.
"Upgrade Failed – An error occurred during the upgrade process. Check logs below.": read the message in the Progress Log. Frequent cause: the images of the new Manager are not on the node.
"Noch nicht möglich, es fehlt: …" when applying: the list names the open prerequisites, for example nodes that are not prepared or missing time synchronisation.
"Auf den Knoten läuft gerade schwere I/O … Hochladen erst danach": wait until the running import has finished.
"Manager nicht erreichbar. Neuer Versuch in wenigen Sekunden …" is expected while the update is being applied. The run continues on the nodes.
The operator update does not check whether a step back is permitted. After installing EasyUC 4.8.7, do not use it for an older Manager version.
After the Manager has been replaced, you may have to log in again because older logins become invalid.