InfoScale™ 9.0 Cluster Server Configuration and Upgrade Guide - Linux

To upgrade InfoScale using YUM

1. Configure an InfoScale repository on the node.

   • Create a repository file (.repo) under /etc/yum.repos.d.

     # cat /etc/yum.repos.d/infoscale<version>.repo

   • Add the following contents to the repository file:

     [repo-infoscale<version>]
     name=Repository for InfoScale infoscale<version>
     baseurl=file:///<image_dir>/rpms/
     enabled=1
     gpgcheck=1
     gpgkey=file:///<image_dir>/rpms/RPM-GPG-KEY-veritas-infoscale

     The values for the baseurl attribute can start with http://, ftp://, or file:///. The URL you choose should be able to access the repodata directory. It also needs to access all the VCS RPMs in the repository that you create or update.

   • Save the changes and exit the file editor.

   • Update the yum repository.

     # yum repoList

     # yum updateinfo

2. If applicable, perform a minor OS version upgrade as follows:

   • Disable the InfoScale repository.

     # yum update --disablerepo=[Infoscale repo]

   • Update the minor OS version.

3. Take one of the following courses of action depending on the InfoScale version from which you are upgrading:

   • Upgrades from InfoScale 7.4.1.x (mixed-version clusters are not supported)

     Run the following command on any node to stop the Cluster Server (VCS) high availability daemon (HAD) across the entire cluster:

     # /opt/VRTSvcs/bin/hastop -all -force

     Note: Do not start HAD on any node until all the nodes have been upgraded to the new version.

   • Upgrades from InfoScale 7.4.2.x onwards (rolling upgrades are supported)

     Run the following command on a single node at a time to forcefully stop HAD:

     # /opt/VRTSvcs/bin/hastop -local -force

4. Upgrade InfoScale.

   Run the following command to update the required rpms:

   # yum update VRTS*

5. Take one of the following courses of action depending on the InfoScale version from which you are upgrading:

   • Upgrades from InfoScale 7.4.1.x (mixed-version clusters are not supported)

     After all the nodes have been upgraded, run the following command on each node to start HAD:

     # /opt/VRTSvcs/bin/hastart

     Note: Do not start HAD on any node until all the nodes in the cluster have been upgraded.

   • Upgrades from InfoScale 7.4.2.x onwards (rolling upgrades are supported)

     Start HAD, if you had stopped it earlier.

     # /opt/VRTSvcs/bin/hastart

6. Manually generate and install the installer scripts for the configuration:

   # /opt/VRTS/install/bin/UXRT<version>/add_install_scripts

7. Reboot the node as follows:

   • If the application service group is online on the current node, fail it over to an alternate node.

     # /opt/VRTSvcs/bin/hagrp -switch <app_service_group> -to <alternate_node_name>

   • Stop the VCS high availability daemon locally.

     # /opt/VRTSvcs/bin/hastop -local

   • Gracefully reboot the node.

     # shutdown -r now

To update the protocol versions and types definition manually

1. Update the VCS protocol version.

   # /opt/VRTSvcs/bin/haclus -version -update <protocol_version>

2. Update the types definition.

   # /opt/VRTSvcs/bin/hatrigger -updatetypes 1 <node_name> 1

3. Update the CFS protocol version.

   # vxupgrade -n <protocol_version> <file_system_name>

4. Update the CVM protocol version.

   # vxdctl upgrade

(Optional - InfoScale 9.0.2 onwards) To update the protocol versions and types definition using the script

• Execute the following script to update the VCS, CVM, and CFS protocol versions and the VCS types definition:

  # python3 /opt/VRTS/install/yum_postupgrade.py