Performing maintenance and updates

Every SC//Reliant™ deployment runs its nodes as two-node clusters, so planned work on a node or on a cluster follows procedures that keep workloads running on the other node and keep the cluster protected. Perform these procedures inside a planned change window, because monitoring checks report alerts while the work is in progress.

• Connecting to a node
• Performing maintenance on a node
• Disabling fencing during maintenance
• Replacing drives or the chassis
• Verifying a node after maintenance
• Updating system packages on a node
• Updating the cluster software, DRBD, and the kernel
• Shutting down and restarting a cluster

Performing maintenance on a node includes the procedures for updating system packages, the cluster software, DRBD, and the kernel. For how maintenance mode and standby work, see Operating the cluster.

In-place upgrades and OS-level changes

Updating an individual system package is not the same as moving a node to a new SC//Reliant platform generation. That larger operation is called an in-place upgrade, and Scale Computing performs it on your behalf. Our engineering teams run the upgrade and coordinate the change window with you in advance. The procedures in this chapter are not a substitute for it, and you should not attempt a generational upgrade yourself.

If your maintenance requires moving to a new platform generation, or you are not sure whether a particular OS-level change falls under an in-place upgrade, contact Scale Computing Support or your account manager before you begin. They can also confirm the release cadence, the staged rollout pattern, and the rollback options that govern platform updates.