Upgrade Guide

Before You Start

1. Purpose: Upgrading is a complex process. Read each item in the upgrade documentation carefully to avoid impacting your business after the upgrade.
2. Applicable products: all versions of FCP-Core, FCP-Suite, and FCP-SE.
3. Make sure to read Version Change Notes. After you confirm the upgrade plan, follow the upgrade steps.

warning

Some components are installed on cluster nodes because they are responsible for task execution. Since the platform cannot accurately determine task runtime status, it will not automatically upgrade these node components during the upgrade process. Therefore, node component upgrades are not supported while the node is still part of a cluster.

Fastone will try to keep old nodes stable, but new features cannot be enabled with old components. Refer to Version Change Notes and plan node component upgrades accordingly.

Upgrade Impact

• Management UI: all platform features will be temporarily unavailable.
• Tasks and jobs: tasks submitted via Application Center before the upgrade will be interrupted. Jobs already submitted via CLI before the upgrade are not affected.
• Cluster management: for the impact of using Ubuntu 22.04 in Fsched clusters, see Operating System.

Pre-Upgrade Checklist

• Management nodes: ensure the Core and Monitor nodes have enough free disk space (about 40 GB) to store the new installation package.
• Task management: ensure there are no submitted tasks still in an unfinished state (specifically tasks submitted via Application Center).
• Cluster management: ensure all clusters are in a stable state (running, powered off, released, or error state).

Perform the Upgrade

See Upgrade Steps.

Post-Upgrade Actions

Cluster Management

1. Fsched versions are backward-compatible, but make sure the head node runs a newer Fsched version than compute nodes. If you need to add nodes after the upgrade, upgrade the head node Fsched to the latest version first.
2. If your Fsched cluster uses Ubuntu 22.04, after the upgrade follow Operating System to reconfigure the cluster as required.

Host Management

After the upgrade, we recommend upgrading all hosts to ensure the services on hosts are up to date.

Cluster Templates

After the upgrade, cluster templates may become unavailable. Edit the template without changing any settings and click Confirm to make the template available again.

Additional Notes

Key Concepts

Cluster: A system composed of multiple compute nodes designed to provide high-performance computing. By connecting multiple computers, a cluster can share resources and process tasks in parallel to improve efficiency and reliability. All workloads are submitted to clusters in FCP, including Fsched clusters and None clusters, to meet different compute needs.

Job: A compute workload in Fsched. In this document, scheduler-submitted compute workloads are referred to as jobs. Jobs are submitted to the cluster and managed by the Fsched scheduler based on resource availability and priority.

Task: A workflow instance submitted via the FCP Application Center. Each task includes one or more Fsched jobs and represents a set of related compute steps. FCP tasks simplify complex scheduler commands and scripts into GUI operations, helping new users get started faster and organize compute workflows more efficiently.