Search Docs...
⌘ K
Before initiating the update process, perform the following checks and complete the following preparations.
Based on the server type of the current virtualization platform, obtain the upgrade ISO file of the target ACOS version in advance.
Only specific versions of AOC can manage ACOS. Therefore, refer to the ACOS Release Notes for the target version first to confirm the compatible AOC version for that cluster. If the current AOC version does not support managing the target version of the cluster, upgrade AOC to a compatible version first.
During the ACOS upgrade process, a CPU instruction set check will be performed, and the CPU of the server should meet the minimum microarchitecture requirements; otherwise, the upgrade process will be terminated. The minimum microarchitecture requirements for the CPU are as follows:
| Architecture | Minimum Microarchitecture Requirement | Notes |
|---|---|---|
| Intel x86_64 | Sandy Bridge | Applicable to physical machines running ACOS in AVE virtualization, and VMware vSphere clusters with EVC configured where SCVM resides in ESXi. |
| AMD x86_64 | Zen | — |
To enhance system performance and address security risks, ACOS periodically provides new kernel versions that are integrated into both the installation and upgrade files of the ACOS new version.
Before proceeding with the upgrade, refer to Kernel update history to verify whether the kernel version used before the upgrade matches the kernel version compatible with the target version. If the versions do not match, update the kernel as needed according to the kernel update records.
If you upgrade the cluster via the upgrade center, you can navigate to the upgrade interface and choose either to upgrade the kernel automatically during the cluster upgrade, or upgrade the kernel separately after the cluster upgrade is complete. For operation details, refer to Arcfra Upgrade Center operation and maintenance.
If you upgrade the cluster via the command line , you need to manually upgrade the kernel after the cluster upgrade is complete. For specific operations, refer to Upgrading the kernel in this document.
Before performing the upgrade, check the memory resources required by the target version of the system. If the available memory does not meet the requirements of the target version, the upgrade will not be able to proceed. In this case, nodes on the AVE platform need to release memory in advance, while SCVMs on the VMware ESXi platform need to expand memory in advance. The specific procedures are as follows:
AVE platform
Nodes can free up memory through the following methods, and after memory is freed up, you can initiate the upgrade process again.
Migrate the virtual machines from the affected node to other nodes with sufficient memory.
If virtual machine migration does not free up enough memory, power off the non-essential virtual machines on the target node.
VMware ESXi platform
Refer to the appendix in Expanding memory for SCVM.
During the cluster pre-upgrade check, if the total data capacity or used data capacity of the current cluster exceeds '80 TiB × the maximum number of hosts allowed by license', the upgrade is not permitted. In this case, apply for and update a license that includes the "Max cluster capacity" to ensure that the licensed capacity is greater than the current total data capacity or used data capacity before proceeding with the upgrade.
If the system partition space is highly utilized, the upgrade components may not operate properly, causing the cluster upgrade to fail. In this case, it is necessary to resolve this issue by cleaning the system partition space before proceeding with the upgrade.
If the cluster has ANS, AKE, or AFS service deployed, hosts may fail to enter into maintenance mode and be unable to upgrade the kernel. The specific causes and solutions are as follows:
ANS or AKE service is deployed in the cluster
If the number of ANS Controller virtual machines, load balancer virtual machines, AKE management cluster control panel node virtual machines, or AKE workload cluster control panel node virtual machines in the cluster is the same as the number of cluster hosts, maintenance mode cannot be entered. This is due to the virtual machine placement group setup that requires these system service virtual machines to reside on separate hosts and prohibits shutting these hosts down.
Solution:
Log in to AOC, navigate to the VM placement group page of the cluster, and then enter the keyword ans or controlplane in the search bar to locate the virtual machine placement groups containing ANS-related system service virtual machines and AKE-related system service virtual machines.
Disable the above-mentioned virtual machine placement groups and proceed with the kernel upgrade.
After the kernel upgrade on all nodes in the cluster is complete, re-enable the placement groups.
AFS service is deployed in the cluster
Because the file controller virtual machine has placement group rules that require it to be on specified hosts and be on different hosts, the file controller virtual machines cannot be migrated to other hosts, preventing hosts from entering into maintenance mode.
For AFS 1.2.1 version
Please contact Arcfra technical support for assistance.
For AFS 1.1.2 version
Go to AOC and take the file controller offline:
In AOC, go to the File storage page, select File storage cluster from the left sidebar, click the name of the target cluster in the list, then click Settings tab and select File controller.
Select a file controller, click the … on the right, and select Take offline.
Perform the secondary confirmation in the dialog box that opens and read the risk warning. Then click Take offline.
Upgrade the kernel on a host that has the file controller disabled.
After upgrading the host kernel, bring the file controller on that host back online:
In AOC, go to the File storage page, select File storage cluster from the left sidebar, click the name of the target cluster in the list, then click Settings tab and select File controller.
Select a file controller, click the … on the right, and select Take online.
Continue performing the above steps on the remaining hosts in the cluster one by one until all hosts complete the kernel upgrade.
Information:
Since taking two or more file controllers offline at the same time may cause the file storage service to become unavailable, AOC restricts the operation to one file controller at a time. Therefore, you need to take file controllers offline one by one and perform the kernel upgrade on the host of each file controller.
For ACOS (AVE) clusters, make sure to disable the dynamic resource scheduler (DRS) before using the upgrade center for upgrading the cluster and the kernel, or manually upgrading the kernel to prevent automatic virtual machine migration during the upgrade, which may affect the upgrade process. To disable the DRS, go to the Overview page of the cluster, then select DRS to disable this feature. You can re-enable it once the upgrade is complete.
Ensure the cluster to be upgraded meets the following requirements:
Connected.