PowerFlex: Upgrade the OS packages on the MVM

• Download PowerFlex Intelligent Catalog (IC) and other files for upgrade

Use this procedure to download the PowerFlex appliance files from the Dell Technologies support site.

1. Copy the /shares/IC/ScaleIO folder to /shares/ScaleIOCurrent to save the current PowerFlex packages. These current packages are used by the Installation Manager during the installation of PowerFlex.
2. On the /shares drive, delete the /share/IC folder to delete the old IC files and accommodate the new files. Ensure 100 GB of disk space on the /shares drive.
3. Create a folder named /share/IC.
4. Log in to the Dell Technologies support site: PowerFlex appliance Intelligent Catalog Software.
5. Click Drivers & Downloads, and filter for the IC files. Select the IC to download.
6. Download the indicated files for a PowerFlex upgrade and any other additional files for your specific IC requirements as specified in the readme file listed. For example, Readme for PowerFlex IC 4x.
    
7. Copy the IC to the jump server shared drive location.

See Dell PowerFlex Appliance with PowerFlex 4.x Upgrade Guide

• Upgrading the OS packages on the management virtual machine

Use the procedures in this section to manually upgrade the embedded operating system VMs running the PowerFlex management platform. Perform each procedure for every management virtual machine in a sequence, repeat until all PowerFlex nodes are updated successfully.

NOTE: Do not perform the updates on more than one node at a time. Doing so may cause damage to the installation of PowerFlex Manager. You cannot access PowerFlex Manager during the management virtual machine upgrade.

Consider the following prior to the manual upgrade:

• For PowerFlex management controller 2.0, perform the OS package upgrades on the management virtual machine prior to RCM 3.7.3.x. From RCM 3.7.3.x or later, the management virtual machine upgrade is part of the resource group upgrade for PowerFlex management controller 2.0 managed by PowerFlex Manager.
• For PowerFlex management controller 1.0, perform the OS package upgrades on the management virtual machine.

• Prepare the management virtual machines for updates

Use this procedure to configure the management virtual machine (MVM) to access and download the OS updates.

Prerequisites

Ensure the following:

• Add and synchronize the latest RCM.
• Record the credentials and hostnames to access the SSH or web console of the management virtual machines.

Steps

1. Browse to https://[ingress-hostname]/httpshare/download/ and authenticate (if prompted).
2. Identify the most recent hash-path based on the timestamp. The timestamp must correlate to when the latest RCM was added.
3. Browse  

   [hash-path]/os/VxFlex4.xSLES15.xRepoX/sles-extras/.

The path contains the PowerFlex release number as well as a repository. The repository is followed by a number. The number depends on how many RCMs are present that contain SLES repositories of the same version. Only one SLES15.3RepoX entry must exist in this path.

For example,

 https://vlan10-ip105.pflexlabs.net/httpshare/download/ 8aaa80b0843444100184346492391acd/os/VxFlex4.xSLES15.xRepoX/sles_extras/

4. Copy the URL path from the address bar.
5. Log in to the management virtual machine.
6. Type 

   sudo zypper addrepo [Use the URL from step 4] <alias>

For example,

sudo zypper addrepo https://[ingress-hostname] /httpshare/download/ 8aaa807c839f3ce20183a439955f108e/os/VxFlex4.xSLES15.xRepoX/sles_extras/  3.7.1.0-pfmp

7. Type the following command to download all applicable packages. This command does not install the packages.

   sudo zypper update -d

The repomd.xml file is unsigned within the custom generated repository. However, the RCM package is signed prior to the release and can be trusted. If prompted to confirm using an unsigned repomd.xml file, enter yes. If prompted to continue, enter y.

8. Repeat steps 5 through 8 for the remaining management virtual machines. Do not install the packages.
9. Verify Packages are available on each MVM 

ls /etc/zypp/repos.d/

• Update the management virtual machines

Use this procedure to update the management virtual machines.

About this task

Use this procedure to perform the following:

• Cordon/drain the management virtual machine node (move running pods to other nodes in the cluster where applicable).
• Install the OS updates and reboot.
• Return the management virtual machine to active status within the cluster with the uncordon command.

Prerequisites

Ensure that you have recorded the credentials and hostnames to access the SSH or web console of the management virtual machines.

Steps

1. Log in to the management virtual machine.
2. Type the following to list all the nodes with the current status:

sudo -i kubectl get nodes

3. Identify the node scheduled for maintenance from the list of nodes and type the following to drain the node. The node is cordoned and is marked as Ready SchedulingDisabled. The scheduler cannot place the new pods on that node.

sudo -i kubectl drain <node> --ignore-daemonsets --delete-emptydir-data

4. Type the following to check for pod redistribution. If any pods are listed in a crash state, wait for five minutes before attempting the command again. Once the command returns an empty result, proceed with the remaining steps. If any pod persists in a Crashloopbackoff state, stop the update and contact Dell Technologies Support for further assistance.

sudo -i kubectl get pods -n powerflex | grep -i crashloop

5. Type the following command to update to apply available updates to the operating system.

sudo zypper -n update

6. If a reboot is required, you are prompted to reboot the node. Type sudo reboot to safely reboot the node while it is cardoned.

It is recommended to reboot the node since most updates impact the running services. A reboot of the VM ensures that all services are restarted.

7. After rebooting, log in to the management virtual machine that is being updated.
8. Type the following command to verify that the node has started all the necessary services. If a connection fails, wait up to five minutes and try again. If the issue persists, contact Dell Technologies Support for further assistance. 

   sudo -i kubectl get nodes

9. Type the following command to return the node to service.

    sudo -i kubectl uncordon <node> 

NOTE: This makes the node available for pods to run. But, does not move the services automatically.

10. Type the following command to verify that the uncordon node is in a ready state.

sudo -i kubectl get nodes 

For example:

Wait for PowerFlex Manager to become available before proceeding to the next management virtual machine in sequence.

11. Type the following command to remove the configuration repository.

    sudo -i zypper removerepo <alias>

    For example, 

    zypper removerepo 3.7.1.0-pfmp.

12. Repeat steps 1 through 10 for the remaining management virtual machines.

See Full Steps in the Upgrade guide 

• Dell PowerFlex Rack with PowerFlex 4.x Upgrade Guide 
• Dell PowerFlex Appliance with PowerFlex 4.x Upgrade Guide