Hi! 

Something more to it.

 

It is important to note, that  vxfs patch must be installed in local zones. It 
means that local zones must be in running or booted state.

If you make failover a zone to another node before upgrade than your local zone 
cannot be in running state on the first node. It can be in installed  state.

But when you start patching  your local zone on the first node , the zone must 
change  its state from "installed" to "booted" to be patched. There is no 
problem if the local zone does not have in its configuration any veritas 
volumes and file systems which have been started up  on the second node. If so  
your local zone  on the first node cannot "see" its resources   and cannot go 
from "installed"  into "booted" state to be patched up. That is the reason why 
"failover resources" must not be present  in local zone configuration. They 
must be mounted from global zone directly.

 

 

Regards, Pavel

 

From: Fred [mailto:mamatec...@mamatec.com] 
Sent: Tuesday, March 27, 2012 12:38 AM
To: Цветков Павел Анатольевич
Cc: colin.y...@gmail.com; veritas-ha@mailman.eng.auburn.edu
Subject: Re: [Veritas-ha] upgrade VCS 5.1 with many zones

 

Hi all my SF-HA origin is 5.1rp1 for Solaris and I should to upgrade to 5.1 SP1 
or 6.0, NOT upgrade Solaris but only SFHA.
What about rolling upgrade, it's possible doing on two node cluster? How to?

Thanks

Fred

2012/3/26 <p.tsvet...@inform.gazprom.ru>

Hello all!

What about using Solaris LiveUpgrade ? It is possible to use it with VCS. There 
are some tricks though ...

You should keep in mind the following.

1. Do not use zpool cluster resource if you don't failover zpools. This is 
because of the fact, that
  Offline entry point for zpool does zpool export. This is a not a good thing 
for "init 6" which must be run after activating New Boot Environment.  If you 
have questions about this point  just ask me.

2. main.cf types.cf must be in synclist file.
3. zfs  file systems must NOT have mountpoint=legacy. Mountpoint must stay 
default.
4. Zone resources . Check DetachZone = false (Zone must stay in installed state 
while it is offlined. This is required to create ABE for zones as well as to 
patch zones).  ForceAttach  must be true if you use Veritas ODM for Oracle DB.
5.  If you have vxfs mounted inside zones mount them from outside. It means 
that vxfs file systems must not be present in zone configuration files , they 
must be mounted with mount command from global zone.
6. You should have the following cluster configuration:




                SapTaskGroup  (monitored from local zone - sap & oracle 
resources without mount points for sap and oracle f.s.)

                    |    online local  firm
                    |


               Mount_SapGroup   (monitored from global zone -  sap & oracle 
mount points on vxfs f.s. - mounted from global zone )

                    |    online local HARD
                    |

               Zone_SapGroup       ( only zone resource itself - monitored from 
local zone)


 If you have dialog instances than you can have them as parallel service groups 
in local zones.



 This is about all.  After that you can do the following:

1. lucreate -n MyNewBoot
2. Upgrade Solaris version with command
 luupgrade -u -n  MyNewBoot   -s /mnt/iso  -k   /var/tmp/sysidcfg


2. Install Recommended Patch Cluster on MyNewBOOT with all zones in parallel 
(check /etc/patches/pdo.conf   num_proc > 1)

   /installcluster -B MyNewBoot --s10patchset

3. install veritas  patches

for i in `cat /sample/patch_order`

do
luupgrade -t -n MyNewBoot  -s /sample  $i
done

4. Luactivate MyNewBoot
5.  init 6



This is not easy to do everything without errors. Please, create your test 
environment and see first.

Good luck!  Pavel




-----Original Message-----
From: veritas-ha-boun...@mailman.eng.auburn.edu 
[mailto:veritas-ha-boun...@mailman.eng.auburn.edu] On Behalf Of Colin Yemm
Sent: Monday, March 26, 2012 5:13 AM
To: Fred
Cc: veritas-ha@mailman.eng.auburn.edu
Subject: Re: [Veritas-ha] upgrade VCS 5.1 with many zones

Short version:

Um, something is wrong. Once you figure out what VCS version you are
running, you should be able to do a rolling upgrade (node-by-node) if
it is 5.1. I haven't been able to find a definitive statement about
5.0 (except how to upgrade from 5.0MP3RP2 to 5.1SP1), but I seem to
remember it can be done, but is a somewhat manual process.

Are you running 5.1 or 5.0?

I ask, because while 5.0 has MP3, the latest VCS version for 5.1 on
UNIX is 51SP1 (actually 51SP1RP2, aka SFHA 5.1 Service Pack 1 Rolling
Patch 2). (There are Point Patches for some packages as well - the
VRTSvxvm on my Clariion-attached servers is 51SP1RP2PP2 ;-)

By my math the next thing to be released on 5.1 will be either
51SP1RP3, or they may just jump straight to 51SP2, depending on how
many bug fixes are included.

Check the patch finder on SORT (sort.symantec.com) for more info.

Colin
Who is having all kinds of fun right now dealing with SFHA versions
going all the way back to 3.5 ;-)

On Sun, Mar 25, 2012 at 4:36 PM, Fred <mamatec...@mamatec.com> wrote:
> Hi all,
> How is the best practice to upgrade VCS 5.1 to 5.1MP3?
>
> the global zone is ZFS, the VCS sg are a lot of zone (with zfs), with
> sap apllication over vxvfs file systems.
> Is it possibile to upgrade one node at time without offline of
> services groups but for example only application?
>
> thanks in advance
>
> Laura
> _______________________________________________
> Veritas-ha maillist  -  Veritas-ha@mailman.eng.auburn.edu
> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha
_______________________________________________
Veritas-ha maillist  -  Veritas-ha@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha

 

_______________________________________________
Veritas-ha maillist  -  Veritas-ha@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha

Reply via email to