Summary of install instructions for 126106-40 and 126107-40

My last blog describe some issues around these patches (please read)
126106-40 Sun Cluster 3.2: CORE patch for Solaris 10
126107-40 Sun Cluster 3.2: CORE patch for Solaris 10_x86
This is a follow up with a summary of best practices 'How to install?' these patches. There is a difference between new installations, 'normal' patching and live upgrade patching.
Important: The mentioned instructions are working if already Solaris Cluster 3.2 1/09 update2 (or Solaris Cluster 3.2 core patch revision -27(sparc) / -28(x86) ) or higher is installed. If running lower version of the Solaris Cluster 3.2 core patch then additional needs are necessary. Please refer to special install instructions of the patches for the additional needs.
Update: 28.Apr 2010
This also apply to the already released -41 and -42 SC core patches, when -40 is not already active

A) In case of new installations:

Install the SC core patch -40 immediately after the installation of the Solaris Cluster 3.2 software.
In brief:
    1.) Install Solaris Cluster 3.2 via JES installer
    2.) Install the SC core patch -40
    3.) Run scinstall
    4.) Do the reboot
Note: Do NOT do a reboot between 1.) and 2.). Follow the EIS Solaris Cluster 3.2 checklist which also has a note for this issue. If not available follow the standard installation process of Sun Cluster 3.2


B) In case of 'normal' patching

It is vital to use the following/right approach in case of patching. Because if you not use the following approach then the Solaris Cluster 3.2 can not boot anymore:
    0.) Only if using AVS 4.0
    # patchadd 12324[67]-05 (Follow Special Install Instructions)
    1.) # boot in non-cluster mode
    2.) # svcadm disable svc:/system/cluster/loaddid
    3.) # svccfg delete svc:/system/cluster/loaddid
    4.) # patchadd 12610[67]-40
    5.) # init 6


C) In case of LU (Live Upgrade feature) to install patches
    1.) Create ABE:
    For zfs root within the same root pool:
    # lucreate -n patchroot
    For ufs on different root drive:
    # prtvtoc /dev/rdsk/c1t3d0s2 | fmthard -s - /dev/rdsk/c1t2d0s2
    # lucreate -c "c1t3d0s0-root" -m /:/dev/dsk/c1t2d0s0:ufs -m /global/.devices/node@2:/dev/dsk/c1t2d0s6:ufs -n "c1t2d0s0-patchroot"
    2.) Install patch into ABE ( patch is already unpacked in /var/tmp )
    # luupgrade -t -n c1t2d0s0-patchroot -s /var/tmp 126106-40
    3.) Activate ABE
    # luactivate patchroot
    4.) # init 6
    # Some errors comes up at this point
    (dependency cycle & ORB error - Please look to example below)
    5.) # init 6 (the second reboot fix the problem) Bug 6938144





My personal recommendation to minimize the risk of the installation for the SC core patch -40 is:
Step 1) Upgrade the Solaris Cluster to
a) Solaris 10 10/09 update 8 and Solaris Cluster 3.2 11/09 update3.
or
b) EIS Baseline 26JAN10 which include the Solaris kernel update 14144[45]-09 and SC core patch -39. If EIS baseline not available use other patchset which include the mentioned patches.
Step 2) After the successful upgrade do a single patch install of the SC core patch -40 by using the installation instruction B) which is mentioned above. In this software state the -40 can be applied 'rolling' to the cluster.

Note: 'Rolling' means: Boot node1 in non-cluster-mode -> install -40 (see B) -> boot node1 back into cluster -> boot node2 in non-cluster-mode -> install -40 (see B) -> boot node2 back into cluster.

Comments:

I had successfully installed the -40 using live update with sun cluster / zones / zfs and global filesystems yesterday.
Althought I had another issue not mentionned here:
After the two reboots, the sysidtool:net was failling the same way as the loaddid was in your blog entry.

I simply booted in non clustered mode and delete the loaddid service and then re-added it from the -40 patch.
The weird thing was that apparently, the loaddid.xml was already in good shape (1.9 version of the SMF startscript and so on)..

Nevertheless, I only had this problem on one node of the cluster, not the other...

Anyway, great article, thanks for it !

Keep up the good work!

Thomas

Posted by Gouverneur Thomas on April 28, 2010 at 05:55 AM CEST #

Post a Comment:
  • HTML Syntax: NOT allowed
About

I'm still mostly blogging around Solaris Cluster and support. Independently if for Sun Microsystems or Oracle. :-)

Search

Archives
« April 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
23
24
25
26
27
28
29
30
   
       
Today