Active IQ Unified Manager Discussions

Splitting an existing Oncommand Core installation

mteeuwen
4,811 Views

Hi,

I want to split an existing OnCommand Core installation in two separate servers:

- One server for Operations Manager (monitoring and reporting) and Protection Manager/Provisioning Manager. This role will be filled by the existing server, where only the Performance Advisor functionality will be disabled.

- One server dedicated for Performance Advisor. This server should get the Performance Advisor history of the existing server.

What is the best scenario in creating the second server for Performance Advisor.

I'm thinking of the following scenario's:

- Clone the data LUN of the existing server and mount the LUN to the new server. Question here is, how do I safely disable the Protection Manager functionality? I'm worried that once I bring up the second server with cloned LUN, the Protection Manager functionality will kick in and start screwing up things because two Protection Manager instances will try to accomplish the same results. How do I prevent this from happening?

- Another possibility is to create a fresh OnCommand installation, register all existing controllers in Operations Manager and then copying the Performance Advisor data from the old server. But how will the existing history data get mapped to the newly discovered controllers in the new server?

What would be the best option in this scenario?

Kind regards,

Marco Teeuwen

5 REPLIES 5

adaikkap
4,746 Views

Hi Marco,

     You have hit the nail on the head. The retiring of Protection Mgr is very critical and needs to be done on the OM Server. Also given that 2 instance of DFM server will be hitting the same set of controller you will have to tone down the monitoring.

Pls refer to the following document for the monitors to turn off for PA only monitoring.

Customizing  OnCommand/DFM server for Performance Advisor (PA) only functionality.

In the mean time pls stay tuned I will give you the detailed steps on how to do this with no impact or downtime.

Regards

adai

adaikkap
4,746 Views

Hi Marco,

     Before I upload the doc I need some more info.

What is the current version of OCUM that this server is Running ?

Is this a Windows or linux server ?

How Many Dataset/Relationship are being managed ?

How many controllers are being monitored ?

Once I get the answers for these question I will upload the Migration doc.

Regards

adai

mteeuwen
4,746 Views

Hi Adai,

5.0 on Windows 2008, I intend to upgrade to 5.1 before the upgrade and to purge the DB.

It monitors about 14 controllers with some 3300 LUNs. PM manages ~4000 SnapVault relationships and ~1200 or so snapmirror relationships.

Rgds, Marco

adaikkap
4,746 Views

Hi Marco,

     Attached is the detailed procedure to do the same. ( Lately I have done this for atleast 4 other large customers)

Terminology:

  1. Current Server: The server currently running all functionality, like Ops-Mgr, Prot-Mgr, Prov-Mgr & Performance Adviser.
  2. PA Only Server: The new server on which we will restore the db created from Current Server and delete all DataSet.
  3. PM Server:The new incarnation of the Current server where Performance Adviser will be disabled and flat files will be deleted.

Pre-Requisite:

  • New OnCommand Core License Key for the PA Only Server
  • 2 Servers, one Linux server for temporary purpose and another windows server for Final PA server

Cavets:

  • One should do this in lab first before doing it in production.So you can estimate how much time this process will take.
  • Also linux is preferred to do all the manipulation as I don't have the equivalent windows cli automation. Any one who can give me the equivalent windows ones are welcomed and much appreciated.
  • Also Linux is faster in doing lot of database related actions that we do.
  • All the manipulation should be done by isolating the linux server from network so that it cant reach the controller and there by no relationships is deleted.

Known and should Ignore:

  • During the process of manipulation,  You will still see job getting spawned and error out. This is fine as we have already isolated the network on the temp controller and they cant reach the controllers.

Order in which Steps  Should be performed.

  1. Current Server: Do steps 2 to 8
  2. Temp Sever:Install OCUM 5.1
  3. Temp Server:Disable/Isolate its network so that it cant reach controller but you can ssh/telnet to the server
  4. Temp Server:Do steps 12 to 31
  5. PA only server:Install OCUM 5.1
  6. PA only server:Do Steps 33 & 34
  7. PM Server(Re-Incarnation of Current Server):Do steps 9 & 10.

Regards

adai

adaikkap
4,746 Views

Hi Marco,

     Forgot to post the xl sheet with commands. Here you go.

Public