Shutting down VCS (phased only)
Shutdown VCS on remaining nodes that are not being upgraded to preventing them from rejoining the cluster.
To shut down the cluster
-
Separate the nodes that are not being upgraded from those that are.
-
Check to see if there are frozen CVM and SFCFS groups, enter
#
/opt/VRTSvcs/bin/hastatus -sum
If the groups are frozen, unfreeze CVM and SFCFS groups using the following commands for each group:
-
Make configuration read/write:
# /opt/VRTSvcs/bin/haconf -makerw
-
Unfreeze the group:
# /opt/VRTSvcs/bin/hagrp -unfreeze
group -persistent
-
Save the configuration:
# /opt/VRTSvcs/bin/haconf -dump -makero
-
Offline the group on each of the remaining nodes of the cluster:
# hagrp -offline
group -sys
system01
# hagrp -offline
group -sys
system02
# hagrp -offline
group -sys
system03
# hagrp -offline
group -sys
system04
where group
is the VCS service group that has the CVMVolDg and CFSMount resource.
Note
Repeat step 2 for each SFCFS service group.
-
Check and repair each file system:
# fsck -F vxfs /dev/vx/dsk/
diskgroup/
volume
Note
Repeat step 3 for each file system.
-
On each node that is not being upgraded, shutdown VCS, enter the following commands:
-
Enter the following commands:
# /opt/VRTSvcs/bin/hastop -local
# /etc/init.d/vxfen stop
# /opt/VRTS/bin/fsclustadm cfsdeinit
-
Check to see if the qlog module is loaded:
# /usr/sbin/modinfo | grep qlog
-
If the qlog module is loaded, then unload it:
# /usr/sbin/modunload -i <module_id>
-
Enter the following commands to stop GAB and LLT:
# /etc/init.d/gab stop
# /etc/init.d/llt.rc stop
-
Proceed to Updating the configuration and confirm startup (phased or full).