Mr Kabootar,
Here's a few items to consider if you're planning on doing a manual
version of clustering.
1. Your master server should be on a virtual hostname as this is the one
that would failover. eg if your 2 systems are PROD1A and PROD1B, then
your Master Server, EMM Server, Storage Units etc should all refer to
PROD1 which will be the failover hostname. While you're at it, if you
associate PROD1 with another IP, it would simplify further and remove
any EMM server, Storage Unit definitions etc that we don't know about,
but integration with VCS does.
2. Your devices should match exactly. Since you're using Solaris, this
is easily accomplished by ensuring your /dev/rmt's and changers are
exactly the same in /etc/devlink.tab.
3. You can used shared disks with VxVM or SVM (Solaris Volume Manager).
4. Both Master Servers should be at exactly the same OS/patch/NBU
versions. It would be best to have the same hardware as well.
5. A manual failover would be the preferred methodology in this scenario.
5. Write a script that will do the failover. ie
i. Shutdown NBU
ii. Unmount the /NBU FS that is on the SAN LUN.
iii. Mount /NBU on the standby system.
iv. Startup NBU
6. Write up a document that's short and precise.
7. Test, solidify script & documentation and retest. Repeat at regular
intervals.
Share your results with the group. We would be interested in knowing
your experience, Mr Bolobaboo Kabootar.
On Sat, Dec 20, 2008 at 5:50 PM, bolobaboo kabootar <bolobaboo at
gmail.com>wrote:
> Ed
> I saw that document and it talks about clustering master server.My
> questions are ..
> 1). Once i put /usr/openv/netbackup folder and its subfolder on SAN lun.
> It will be protected. Once stand by server is up linking
> /usr/openv/netbackup on new system will work ?
> 2). Do netackup puts any other files on Solaris system folder ? May be i
> should install both primary and stanby with same release and patch. Will
> that take care of any OS related files ?
>
It's not that simple if your master server is also a media server. You need
to take into account EMM, the device configurations, and probably other
stuff that I haven't thought of.
If you have common SAN storage, then clustering really is the way to go...
> Thank you for your suggestion.
>
> On Sat, Dec 20, 2008 at 6:00 PM, Ed Wilts <ewilts at ewilts.org> wrote:
>
>> 2008/12/20 bolobaboo kabootar <bolobaboo at gmail.com>
>>>
>>> I have one master server 6.5.3 on solaris 10. We are not planning to
>>> cluster it. I was thinking is there way to synch existing master
server with
>>> another stand by solaris system. Then bring this stand by server if
anything
>>> goes wrong with this pimary master server. This way will have
minimum down
>>> time.
>>> Does any body has implemented this approch ? If you have done
this can
>>> you passs on steps ?
>>>
>>
>> There is a document on the Symantec web site on implementing highly
>> available NetBackup solutions. Start by reading that...
>>
>
>
.../Ed
Ed Wilts, RHCE, BCFP, BCSD, SCSP, SCSE
ewilts at ewilts.org
--
Larry Mascarenhas
LMascare AT comcast DOT net
_______________________________________________
Veritas-bu maillist - Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
|