Upgrade impact | PRA
ImportantIf your on-premises appliances are in an Atlas cluster or a failover pair, upgrading to Base Software 8.4 turns off automatic updates on every node. After the upgrade, they stay off until you turn them back on.
What changed in Base Software 8.4
In Base Software 8.3 and earlier, the setting Automatically install the latest updates during maintenance day(s) and time installs only patches on an appliance in an Atlas cluster. In Base Software 8.4, the same setting also installs full product upgrades, which the primary node coordinates across the cluster.
Because this setting now does more than it did when you turned it on, the upgrade turns it off. You can then decide whether to turn it back on with the new behavior.
Who is affected
The change applies only to on-premises appliances that are in an Atlas cluster or a failover pair when they upgrade to Base Software 8.4.
The following are not affected:
- Pathfinder deployment
- Cloud deployments
- FedRAMP deployments
- Standalone appliances that are not clustered and not in a failover pair
Settings turned off by the upgrade
The upgrade turns off the following settings on every node, including the primary node:
- Automatically install the latest updates during maintenance day(s) and time
- Install updates on this node together with the other nodes. This setting is new in Base Software 8.4. The primary node does not display it, because the primary node always coordinates updates.
The setting Automatically install disruptive critical updates during the Maintenance Window is not changed.
What still installs automatically
Critical security updates continue to install automatically:
- Non-disruptive critical updates install as soon as they are eligible.
- Disruptive critical updates install during the maintenance window, if Automatically install disruptive critical updates during the Maintenance Window is turned on.
Other patches and product upgrades do not install automatically until you turn automatic updates back on.
Maintenance window settings no longer follow the primary node
A node that does not take part in cluster updates manages its own updates. It no longer takes the Maintenance Window Start Time or Maintenance day(s) settings from the primary node. The same applies to the automatic updates setting, the update channel, the disruptive critical updates setting, and the minimum days notice.
Each node keeps its current values, so you do not see a change immediately after the upgrade. However, if you later change Maintenance Window Start Time or Maintenance day(s) on the primary node, the other nodes keep their old values until you turn Install updates on this node together with the other nodes back on.
Turn automatic updates back on
- On the primary node, go to /appliance > Updates.
- In the Maintenance Window section, check Automatically install the latest updates during maintenance day(s) and time, choose an Automatic Updates Channel, check the maintenance window and days, and then click Save.
- On each of the other nodes, go to /appliance > Updates.
- In the Cluster Updates section, check Install updates on this node together with the other nodes.
- Confirm the message, and then click Save. The node might start installing updates within minutes to catch up with the rest of the cluster.
- section, click Resync now, wait a moment, and then reload the page
- On the primary node, return to the Cluster Updates section, click Resync now, wait a moment, and then reload the page. Confirm that the Coordination column shows Participating for each node. The message No other node is participating, so each node has to be updated manually no longer appears.
If you want to keep some nodes on their current version while you test an upgrade, leave participation and automatic updates turned off on those nodes. For more information, see Upgrade a cluster in stages.
On a failover pair, automatic product upgrades remain on hold while Enable Automatic Failover is turned on for the backup node (/login > Management > Failover), even after you turn automatic updates back on. Patches are not held. For more information, see How cluster updates run.
Updated about 23 hours ago