Here the root filesystem is mirrored one using SVM. So prior to OS patching,we should break the mirror for back-out purpose. After the OS patching ,if the system is not coming up in new patch level,then we have a valid mirror disk to boot the system in older configuration.
Another think you need to make sure that VCS shouldn’t start in the OS start up.If its starts then it will try to bring up the cluster resource and it will kill your time. So better to not place the VCS scripts in the start-up.
We will see high level OS patching step by step.
1.Copy the OS bundle to /var/tmp and unzip it .
2.Check the installed zone
test1#zoneadm list –cv
ID NAME STATUS PATH BRAND IP
0 global running / native shared
6 zone1 running /export/zones1 solaris8 shared
7 zone2 running /export/zones2 native shared
10 zone3 running /export/zones3 native shared
Assume,
· zone1,zone2 are in VCS control
· zone3 is not in VCS control.
3.Change the cluster ID for the cluster to run in single node mode./
# cp –p /etc/llttab /etc/llttab.before_Patching
# vi /etc/llttab
…
set-cluster 953
…
4.Take the necessary configuration backup of server
(df -h,vxdg list,vxprint -hvt,netstat -rn,metatstat,metastat -p)
4. Split the mirror
– Split the mirror and then clear the detached sub-mirrors.
– Remove the Meta-State DB replicas.
5. Shutdown the VCS
# hastop -local
The local zones that are not controlled by VCS, do:
Log into the local zones, and shut them down nicely.
test1# zlogin -C local-zone-name
local-zone-name# shutdown -i5 -g0 –y
Disable VCS automatic start
test1#cd /etc/rc2.d
test1#mv S71llt _S71llt
test1#mv S95gab _S95gab
test1#cd /etc/rc3.d
test1#mv S99vcs _S99vcs
6.#import all the zones disk group manually and start the volumes
Bring the Global (test1) server down to ”’single-user”’ mode.
test1# reboot — -s
7. mount all the zone root manually
test1#mount –F vxfs /dev/vx/dsk/zone1-root/vol1 /export/zones1
test1#mount –F vxfs /dev/vx/dsk/zone2-root/vol2 /export/zones2
test1#mount –F vxfs /dev/vx/dsk/zone3-root/vol3 /export/zones3
8. Install the PatchBundle:
test1# cd /var/tmp/10_Recommended
test1# ./installcluster –s10cluster
Now you need to reboot before installing the Veritas Patches Cluster
test1# sync;sync;sync
test1#reboot — -s
Make sure the server has booted properly.
test1# svcs -xv (verify the svcs output)
9.Mount all the zone root manually again
test1#mount –F vxfs /dev/vx/dsk/zone-root-zone1/vol01 /export/zones1
test1#mount –F vxfs /dev/vx/dsk/zone-root-zone2/vol02 /export/zones2
test1#mount –F vxfs /dev/vx/dsk/zone-root-zone3/vol01 /export/zones3
10.For zone1 & zone2 which are in VCS control,
zoneadm -z zone_name attach -u
zoneadm -z zone_name boot
# install HBA patch on global test1
# install Veritas MP3RP4 patch on global test1
./installrp test1
– make sure MP3RP4 is installed
#pkginfo –l VRTSvcs
11.Put back VCS start up script
test1#cp –p /etc/llttab.before_Patching /etc/llttab
test1#cd /etc/rc2.d
test1#mv _S71llt S71llt
test1#mv _S95gab _S95gab
test1#cd /etc/rc3.d
test1#mv _S99vcs S99vcs
Reboot the machine
init 6
Re-enable the SVM mirror again.
Thank you for reading this article.Please leave a comment if you have any doubt ,i will get back to you as soon as possible.
Rohtash says
So interesting & helpful for providing knowledge about VCs node patching and also helpful for zones process during patch.