IBM Spectrum Fusion

 View Only

Compute node firmware management made easy and seamless on IBM Spectrum Fusion HCI

By Pruthvi T D posted Thu November 03, 2022 05:41 AM

  

Overview:

IBM Spectrum Fusion HCI 2.3 was made globally available from September 2022 with some exciting new features to manage the compute resources. Firmware management of a node is one such new feature.

Why upgrade firmware?

It is always recommended to keep the firmwares up to date to enhance the functionality, resolve issues, to improve security, compatibility, and for higher system uptime.

How is it made easy?

As it is recommended to be at the latest level of firmware version, compute firmware version is continuously monitored. Whenever there is newer version available, it will be notified to the users and they can perform the upgrade by a single-click.

How is it made seamless?

It is not required for the user to know the specifics of the firmware corresponding to the hardware component to download, since it is bundled as part of the IBM Spectrum Fusion HCI software.

As part of the Fusion HCI firmware upgrade process, tasks like migrating the workload, rebooting the node post upgrade for the new firmware to take effect, making the node back to available state will be taken care. Hence, the user is relieved from performing all these tasks manually.

Background:

The following is the list of different compute hardware models that are currently supported in IBM Spectrum Fusion HCI appliance:

  • Lenovo SR630 – Intel Cascade Lake machine type: 7X02; Model: F01
  • Lenovo SR645 – AMD CPU machine type: 7D2XC; Model: C00, C01 (9155-C00, 9155-C01, 9155-C04, 9155-C05)
    • Storage intensive servers with NVMe disk drives for IBM Spectrum Scale ECE(Erasure code edition).
    • Compute intensive servers without NVMe drives.
  • Lenovo SR665 – AMD CPU machine type: 7D2VC; Model: G01

Following is the list of firmware and device drivers that are upgraded as a part of HCI Fusion’s firmware upgrade:

  • System firmwares – Baseboard management controller (BMC), Unified Extensible Firmware Interface (UEFI)
  • PCIe (peripheral component interconnect express) adapters – Ethernet adapters, redundant array of independent disks (RAID) controllers
  • Firmware for hard drive (serial advanced technology attachment - SATA, nonvolatile memory express - NVMe)
  • Device drivers in the operating system - Lenovo XClarity Provisioning Manager (LXPM)

How can user upgrade the Fusion HCI compute node firmware?

New IBM Spectrum Fusion HCI software bundle includes latest version of compute firmware. Following the IBM Spectrum Fusion software update, the user interface notifies the user with a notification banner stating the availability of the newer compute firmware version. User can perform compute firmware upgrade whenever he plans to do so. This notification disappears after updating the firmware on all the HCI bare-metal hosts of the cluster to the version supported by the current IBM Spectrum Fusion HCI software version.

Img-1: Firmware upgrade notification banner

  1. From the list of nodes, user can select one or more nodes for firmware upgrade.
  2. In case user has selected multiple nodes for upgrade, upgrade happens sequentially and not necessarily in the selected order.
  3. The firmware upgrade on a single node would approximately take an hour.
  4. In the event of update failure, user has the option to select retry upgrade on the failed node.
  5. As and when the upgrade progresses, the node goes through following changes:
    1. Migration of all the workloads from the target to other nodes, node will go into maintenance mode.
    2. Firmware upgrade happens.
    3. Node gets rebooted
    4. Node joins the IBM Spectrum Scale Cluster again and is available for workload consumption.
  6. In the event of firmware upgrade failure, user has the option to retry the upgrade.
  7. During the upgrade progress, user has an option to select/add new nodes for upgrade.
  8. During the upgrade progress, firmware upgrade can be cancelled only on the pending nodes and cannot be cancelled on the node that is getting upgraded.
  9. Upon successful completion of firmware upgrade, the upgraded firmware version gets displayed under “Firmware” column on the nodes list page.

 Following five images capture the firmware upgrade process at different phases.

Img-2: Firmware upgrade confirmation widget
Img-3: Firmware upgrade in progress

 

Img-4: Firmware upgrade scheduled(node in upgrade schedule pool)

 

Img-5: Cancel Firmware upgrade on failed node
Img-6: Cancel failed and scheduled firmware upgrade

 
Conclusion:

IBM Spectrum Fusion HCI appliance firmware can be updated without the need for the following tasks:

  • To download the firmware binaries for different hardware components.
  • To boot up a compatible OS on the target node.
All this is performed without the workload downtime in the container environment. Also there are no additional steps involved in upgrading cluster from enterprise registry or a cluster with internet access.

References:

  1. https://lenovopress.lenovo.com/lp0656-lenovo-thinksystem-firmware-and-driver-update-best-practices
  2. Compute Firmware Upgrade events and error codes: https://www.ibm.com/docs/en/spectrum-fusion/2.3?topic=codes-bmyup2106
  3. Reference link to KC : https://www.ibm.com/docs/en/spectrum-fusion/2.3?topic=nodes-upgrading-node-firmware

Acknowledgement

Sincere thanks to Sanjay Verma, Shajeer Mohammed and Shyamala Rajagopalan for helping review the article.

 


#Featured-area-2
#Featured-area-2-home
#Highlights
#Highlights-home
0 comments
508 views

Permalink