StoreVirtual Storage
1748181 Members
4058 Online
108759 Solutions
New Discussion юеВ

Re: Upgrade 8.5 to 9 with Two Site

 
SOLVED
Go to solution
Fableo
Advisor

Upgrade 8.5 to 9 with Two Site

 

Hi,

 

I have 6-node P4500. 3 on Site A and 3 on site B.

The six nodes are used with a Windows Cluster HyperV with windows 2008R2 SP1
I would like to do the upgrade without interruption.

How do I do?

 

Thanks  :smileyhappy:

 

11 REPLIES 11
Bart_Heungens
Honored Contributor

Re: Upgrade 8.5 to 9 with Two Site

Hi,

 

Which version are you running now? Is it a single or a multi site?

 

As such, normally you can do the upgrade without downtime if you do it node per node, and see that you have quorum, which means that you have a majority of managers running. In your case this is 4...

 

First update the CMC to version 9.0.1 and than the nodes from within the CMC...

 

Kr,

Bart

--------------------------------------------------------------------------------
If my post was useful, clik on my KUDOS! "White Star" !
Fableo
Advisor

Re: Upgrade 8.5 to 9 with Two Site

I have two Management Group. The first "Mgt Exchange" node has two (one in each cluster). A node is dedicated to an Exchange 2010 server using the BI DAG.
I launched the update and I have not had the opportunity to select a node. The updating is done simultaneously for both. So stop servers.

I have a second Management Group "Mgt Cegos" which has a 6 node cluster. Three node on site A and Node Three on the site B. They are 8.5.00.0370.0 version. I want to be sure that the update will be seamless because the 6 nodes are used for HyperV Cluster with 80 virtual machine.


thank you

Bart_Heungens
Honored Contributor

Re: Upgrade 8.5 to 9 with Two Site

You always must be sure that you have a quorum, or majority of managers running... So even with only 2 nodes in a cluster, if you have a FOM or Virtual Manager running, there isn't downtime on volume level...

 

The update is indeed done simultaneously in 1 job, but in the background he will do 1 node at a time... So there shouldn't be downtime...

 

In case of the 6-node cluster, same story... See that you have a majority of manager running. In this case normally you have 5 managers running (4 on the nodes and a fifth manager being a FOM or Virtual Manager). So see that you have at least 3 or 4 managers running...

The automatic update process will do this in the background anyway... Did it already several times, never had issues...

 

Good luck!

--------------------------------------------------------------------------------
If my post was useful, clik on my KUDOS! "White Star" !
Fableo
Advisor

Re: Upgrade 8.5 to 9 with Two Site

I do not understand.
You talk of a quorum HyperV or P4500?
I have 5 in the configuration manager on the Cluster Storage System 6.
attached a screenshot.

Bart_Heungens
Honored Contributor

Re: Upgrade 8.5 to 9 with Two Site

Hi,

When I talked about the quorum, in my case I was talking about quorum in the Lefthand environment... Sorry for not stating clear...

OK I see you have 5 managers running  on the nodes itself, which is also OK off course...

 

However, since you have 3 nodes on A site and 3 nodes on site B, in case of a network problem between both sites, this means that you will loose access to all volumes in both sites since you don't have a majority or Lefthand quorum at that moment... Per site you can have a maximum of 3 managers which is not a majority for 6 nodes... You will need a fouth one... And that is possible by means of that so called FOM Failover Manager...

 

Now for the update is this not an issue, the update will be done node per node anyway so in that case enough managers will be there to keep the volumes online...

 

For High Availability reasons, I would suggest a FOM functionallity...

 

 

Kr,

Bart

--------------------------------------------------------------------------------
If my post was useful, clik on my KUDOS! "White Star" !
Fableo
Advisor

Re: Upgrade 8.5 to 9 with Two Site

ok, thank you. :smileyhappy:
I added the last node as a manager. Now I have three manager  at the site A and three manager at the site B.
I will start the update tonight.

Concerning the FOM, it brings me what could be more environmental HyperV? :smileyindifferent:

Bart_Heungens
Honored Contributor

Re: Upgrade 8.5 to 9 with Two Site

Hi,

 

Best practice from HP is to have an unpair amount of managers (3 or 5), and a maximum of 5...

 

In case of 6 managers (3 on both sites), you will loose I/O to your volumes in case of a split-brain situation...

 

So better is to have in 1 site 3 managers and in the other site 2 managers active. In case of a split-brain situation, the site with 3 managers will stay 'alive', the other site with 2 managers will loose I/O at that moment...

 

If site A (with the 3 managers) explode (or whatever other disaster), site B will however loose I/O since it does not have a majority of managers at the moment of the disaster. A manual intervention is needed at that moment...

 

And that is why the FOM can be an interesting piece in the Lefthand network when it is installed in a third location and can decide for you which site stays alive...

 

It may look complex but it isn't at all... But in my trainings regarding Lefthand, this is always a hot topic of discussion since it is really scalable...

--------------------------------------------------------------------------------
If my post was useful, clik on my KUDOS! "White Star" !
Fableo
Advisor

Re: Upgrade 8.5 to 9 with Two Site

Thanks for the reply.
I am back to 5 managers (3 on A site and 2 site B)
aaguilos
Occasional Visitor

Re: Upgrade 8.5 to 9 with Two Site

You just have to make sure that the volume have data protection level of RAID10. As far as DSM goes, I had to fail over our SQL cluster from one node to the other and install them one at time. 

 

Also, prior to the upgrade you may want to increase the bandwidth priority. Since I did our upgrade during our least traffic time, I bump it up to max to make the whole restriping faster.

 

I used this link during the upgrade to make sure everything is prepped and ready:

 

http://bizsupport1.austin.hp.com/bc/docs/support/SupportManual/c01886197/c01886197.pdf