VCS Upgrade

Steps to take to perform SFHA upgrade from 5.0  to 5.1  using phased minimal downtime procedure.

This procedure shows how to upgrade a 4 node cluster from 5.0  to 5.1 . All four nodes will be upgraded for VCS together at the beginning of this procedure. Then Storage Foundation can be upgraded either one node or two nodes at a time, while the VCS Service Groups are running on the other two nodes. This procedure is not recommended for CFS/CVM environment.

I . 5.1SP1 VCS upgrade needs to be done first on all of the cluster nodes:
1. Persistent freeze service groups on all nodes
Commands for doing this:
– Make the cluster configuration read / write.
# haconf -makerw
– Persistent freeze each and all cluster systems (perform this step for each cluster node).
# hasys -freeze -persistent <node>
– Save and dump the cluster configuration.
# haconf -dump -makero
2. Now stop the VCS services – but keep processes and databases running outside of VCS:
# hastop –all -force
3. Run ./installvcs from installation software, and select VCS only.
– Select ‘Veritas Cluster Server’ to install
– Do you agree with the terms of the End User License Agreement as specified in the EULA.pdf file present on media? [y,n,q,?] y
– Select default ‘2’ for what to install for VCS
– enter all node names when prompted.
– Answer ‘y’ to continue this installation.
4. Now start VCS processes on all 4 nodes by running the following on all four nodes:
# ./installvcs –start <node1> <node2> <node3> < node4>
5. Now upgrade VCS to 5.1SP1RP1:
# cd <5.1SP1RP1 installation software path>
# ./installrp -upgrade_nonkernelpkgs <node1> <node2> <node3> <node 4>
6. After completion of previous step persistent unfreeze service groups all nodes
Commands for doing this:
– Make the cluster configuration read / write.
# haconf -makerw
– Persistent unfreeze each and all cluster systems (perform this step for each cluster node).
# hasys -unfreeze -persistent <node>
– Save and dump the cluster configuration
# haconf –dump -makero

II. PHASE 1: Now prepare cluster for phased upgrade for Storage Foundation 5.1SP1:
1. Switch all service groups from two nodes (to be upgraded in next step) to other two nodes.
– Make the cluster configuration read / write.
# haconf –makerw
FOR PHASE 2 OF THIS PROCEDURE, THE CLUSTER NODES WILL NEED ‘PERSISTENT UNFREEZE’:

# hasys -unfreeze -persistent <node>
Now switch the service groups to two nodes:
# hagrp -switch <service_group> -to <node>
2. Persistent freeze all nodes
Commands for doing this:
– Persistent freeze each and all cluster systems (perform this step for each cluster node).
# hasys -freeze -persistent <node>
– Save and dump the cluster configuration.
# haconf -dump -makero

III. Upgrade Storage Foundation on one or two nodes at a time to Storage Foundation 5.1SP1:
1. Run ./installer to install Storage Foundation on nodes not having any service groups running
– Select Storage Foundation to install (option 2)
– Do you agree with the terms of the End User License Agreement as specified in the EULA.pdf file present on media? [y,n,q,?] y
– Select software to install, option 3 for all Storage Foundation product
– Enter node names installing Storage Foundation
– Answer ‘y’ to continue this installation.
2. Check for presence of /etc/vx/reconfig.d/state.d/install-db
If this file is present, delete it – this will prevent vxvm from starting.
3. Once the Storage Foundation 5.1SP1 installation has completed, reboot the newly upgraded systems.

IV. Upgrade the same node or two nodes to Storage Foundation 5.1SP1RP1:
1. Use patchadd command to install the following patches for RP1:
NOTE: The following patches are for Solaris 10.
VRTSvxfs: 142634-07
VRTSvxvm: 142629-09
VRTSsfmh: 144159-01
VRTSodm: 143271-06
OR to use ‘installrp’:
# ./installrp -upgrade_kernelpkgs <node>
NOTE: Use of ‘installrp’ will upgrade VCS and SF! This is not recommended for this procedure as this will cause the Cluster nodes to be at different versions of the product.
If wanting to upgrade SFHA to 5.1SP1RP1 – take total downtime to do this so all the cluster nodes will be upgraded at the same time.
V. Upgrade the same node(s) to Volume Manager 5.1SP1RP1P1:
1 Install patch 142629-10
# patchadd 142629-10
2 Check for presence of /etc/vx/reconfig.d/state.d/install-db
If this file is present, delete it – this will prevent vxvm from starting.
3 Reboot the node(s) after getting the above patches for both RP1 and P1 installed on the node(s) and join them back into the cluster.

Repeat Steps II. – V. for the rest of cluster nodes for Phase 2.

VI. After all nodes of the cluster are at 5.1SP1RP1P1 for Storage Foundation – unfreeze service groups, and switch to the cluster nodes where they are supposed to be running:
1. Make the cluster configuration read / write.
# haconf -makerw
2. Persistent unfreeze the cluster nodes:
# hasys -unfreeze -persistent <node>
3. Switch the service groups to the cluster nodes where they belong:
# hagrp -switch <service_group> -to <node>
4. Save and dump the cluster configuration
# haconf –dump -makero

VCS upgrade completed succesfully