Updating the configuration and confirming startup

Perform the following steps on each upgraded node.

To update the configuration and confirm startup

  1. Remove the /etc/VRTSvcs/conf/config/.stale file, if it exists.
    # rm -f /etc/VRTSvcs/conf/config/.stale
  2. Reboot the upgraded nodes, if root disk is encapsulated.
    # /usr/sbin/shutdown -i6 -g0 -y
  3. After the nodes reboot, verify that LLT is running:
    # lltconfig
      LLT is running
  4. Verify GAB is configured:
    # gabconfig -l | grep 'Driver.state' | \
      grep Configured
      Driver state : Configured
  5. Verify VxVM daemon is started and enabled:
    # /opt/VRTS/bin/vxdctl mode
      mode: enabled
  6. Confirm all upgraded nodes are in a running state.
    # gabconfig -a
  7. If any process fails to start after the upgrade, enter the following to start it:
    # /opt/VRTS/install/installsfcfsha<version> -start sys1 sys2

    Where <version> is the specific release version.

    See About the Veritas installer.

  8. After the configuration is complete, the CVM and SFCFSHA groups may come up frozen. To find out the frozen CVM and SFCFSHA groups, enter the following command:
    # /opt/VRTS/bin/hastatus -sum

    If the groups are frozen, unfreeze CVM and SFCFSHA groups using the following commands for each group:

    • Make the configuration read/write.

      # /opt/VRTS/bin/haconf -makerw
    • Unfreeze the group.

      # /opt/VRTS/bin/hagrp -unfreeze group_name -persistent
    • Save the configuration.

      # /opt/VRTS/bin/haconf -dump -makero
  9. If VVR is configured, and the CVM and SFCFSHA groups are offline, bring the groups online in the following order:

    Bring online the CVM groups on all systems.

    # /opt/VRTS/bin/hagrp -online group_name -sys sys1
    # /opt/VRTS/bin/hagrp -online group_name -sys sys2

    where group_name is the VCS service group that has the CVMVolDg resource.

    Bring online the RVGShared groups and the virtual IP on the master node using the following commands:

    # hagrp -online RVGShared -sys masterhost
    # hares -online ip_name -sys masterhost

    Bring online the SFCFSHA groups on all systems.

    # /opt/VRTS/bin/hagrp -online group_name -sys sys1
    # /opt/VRTS/bin/hagrp -online group_name -sys sys2

    where group_name is the VCS service group that has the CFSMount resource.

    If the SFCFSHA service groups do not come online then your file system could be dirty.

    Note:

    If you upgrade to SFCFSHA 6.0.1 and the file systems are dirty, you have to deport the shared disk group and import it as non-shared. After the import, run fsck. fsck should succeed. Then deport the disk group and import it back as shared.

  10. Find out which node is the CVM master. Enter the following:
    # vxdctl -c mode 
  11. On the CVM master node, upgrade the CVM protocol. Enter the following:
    # vxdctl upgrade