1748089 Members
4744 Online
108758 Solutions
New Discussion

Re: Cluster Migration

 
SOLVED
Go to solution
Mehul5
Frequent Advisor

Cluster Migration

Hello Everyone,

 

We are planning to migrate one entire node which is in cluster from to a new hardware.

 

All the parameters such as ip address,hostname,heart beat ip,etc will remain same.

 

We have booted the new server with the DRD disk and now we want to move the cluster on this server.

 

In this SgeSAP should also be installed.

 

Can anyone please tell me the POA and what are the pre-requisites for it.

 

Regards,

Mehul

8 REPLIES 8
pooderbill
Valued Contributor

Re: Cluster Migration

>> We are planning to migrate one entire node which is in cluster from to a new hardware.

Not very helpful. Infomation needed:

 

What models of hardware are in the current cluster including the one being replaced?

What model is the new hardware?

 

>> We have booted the new server with the DRD disk and now we want to move the cluster on this server.

So the new server is already running?How was it installed m(Ignite, DVD cold install)?

Does it have SG software installed? Is it the same version as the running cluster?

 

Mehul5
Frequent Advisor

Re: Cluster Migration

Hi Pooderbill,

 

I am sorry for the incomplete information.

 

We have 2 rx8640 servers both of which are in cluster and both have been configured when the fresh installation was done so I am not sure how initially the y were installed.

 

Both are having SGeSAP version B.5.00 and serviceguard version as A.11.18 and OS version of March 2009.

 

Both the nodes are on separate servers(i.e one node on 1 rx8640).

 

We have to move one of the node from rx8640 to BL890ci2 while the other(failover) node will be on same rx8640.

 

This new server has OS of March 2012 and no SGeSAP is installed and service guard version as A.11.20.

 

Kindly tell if anything more is required.

 

Regards,

Mehul

 

 

Bill Hassell
Honored Contributor

Re: Cluster Migration

>> serviceguard version as A.11.18 and OS version of March 2009

 

ServiceGuard 11.18 is incompatible with SG 11.20, and the OS (Mar 2009) may need extensive patching to bring it up to to the OS patch on the blade. There is a *lot* of work necessary to accomplish this, especially if this is a production system. And the hardware is very different between the two systems - not a recommended configuration. You will save a lot of sysadmin time (days, perhaps weeks) and also save a lot of package downtime by purchasing another blade, putting the latest patches on the two blades and developing the SG environment between the two blades. Then migrate the package(s) over to the new cluster.



Bill Hassell, sysadmin
Mehul5
Frequent Advisor

Re: Cluster Migration

Hi Bill, 

 

Thanks for the information but we need to migrate on an urgent basis and the same has been approved by the customer also.

 

Even though this is not a recommended activity, the customer still wants to go ahead.

 

And yes this is a production system and the heart of this site.

 

Regards,

Mehul

Bill Hassell
Honored Contributor

Re: Cluster Migration

>> but we need to migrate on an urgent basis

 

This doesn't change the migration steps.

 

>> Even though this is not a recommended activity, the customer still wants to go ahead.

 

Still won't change the requirements or results.

 

>> And yes this is a production system and the heart of this site.

 

All the more reason to perform the required steps (patch both systems, upgrade Service Guard to 11.20, etc)

Otherwise, all the money spent to purchase Service Guard will be wasted as it won't start or failover.

 

If you perform all the patching and SG migration, then you should be able to configure the blade to work with Service Guard.

 

NOTE: There will be mandatory production downtime.

If this is not possible, then abandon the project and run the two systems independently.

 



Bill Hassell, sysadmin
Mehul5
Frequent Advisor

Re: Cluster Migration

Hi Bill,

 

Thnax for the reply.

 

I understand the consequences and risks involved in going forward with this activity.

 

Stll can  you please list out the necessary important steps to carry ot this activity.

 

Regards,

Mehul

Bill Hassell
Honored Contributor
Solution

Re: Cluster Migration

There are no simple 1-2-3 steps to accomplish this task so I will point you to the reference manuals that will explain the steps you need to match your situiation.

 

SG 11.18 manuals:

http://h20565.www2.hp.com/portal/site/hpsc/template.PAGE/action.process/public/psi/manualsDisplay/?sp4ts.oid=4162060&javax.portlet.action=true&spf_p.tpst=psiContentDisplay&javax.portlet.begCacheTok=com.vignette.cachetoken&spf_p.prp_psiContentDisplay=wsrp-interactionState%3DdocId%253Demr_na-c02056214%257CdocLocale%253Den_US&javax.portlet.endCacheTok=com.vignette.cachetoken

 http://h20565.www2.hp.com/portal/site/hpsc/template.PAGE/action.process/public/psi/manualsDisplay/?sp4ts.oid=4162060&javax.portlet.action=true&spf_p.tpst=psiContentDisplay&javax.portlet.begCacheTok=com.vignette.cachetoken&spf_p.prp_psiContentDisplay=wsrp-interactionState%3DdocId%253Demr_na-c02029970%257CdocLocale%253Den_US&javax.portlet.endCacheTok=com.vignette.cachetoken

 

SG 11.20 manuals:

http://h20565.www2.hp.com/portal/site/hpsc/template.PAGE/action.process/public/psi/manualsDisplay/?sp4ts.oid=4162060&javax.portlet.action=true&spf_p.tpst=psiContentDisplay&javax.portlet.begCacheTok=com.vignette.cachetoken&spf_p.prp_psiContentDisplay=wsrp-interactionState%3DdocId%253Demr_na-c03942313%257CdocLocale%253Den_US&javax.portlet.endCacheTok=com.vignette.cachetoken

 http://h20565.www2.hp.com/portal/site/hpsc/template.PAGE/action.process/public/psi/manualsDisplay/?sp4ts.oid=4162060&javax.portlet.action=true&spf_p.tpst=psiContentDisplay&javax.portlet.begCacheTok=com.vignette.cachetoken&spf_p.prp_psiContentDisplay=wsrp-interactionState%3DdocId%253Demr_na-c03942321%257CdocLocale%253Den_US&javax.portlet.endCacheTok=com.vignette.cachetoken

 

 

You mentioned DRD but with no explnation. Was this a DRD clone on an rx8640? If so, have you booted the rx8640 to run on the clone disk? I can only assume that you are running on the rx8640 DRD disk so it will be easy to reboot back to the original disk, correct?

 

DRD manual:

http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=4&cad=rja&uact=8&ved=0CDkQFjAD&url=http%3A%2F%2Fh20564.www2.hp.com%2Fportal%2Fsite%2Fhpsc%2Ftemplate.PAGE%2Faction.process%2Fpublic%2Fkb%2FdocDisplay%2F%3Fjavax.portlet.action%3Dtrue%26spf_p.tpst%3DkbDocDisplay%26javax.portlet.begCacheTok%3Dcom.vignette.cachetoken%26javax.portlet.endCacheTok%3Dcom.vignette.cachetoken%26docId%3Demr_na-c02...

 

 

On the blade, copy the /etc/cmcluster directory from one of the rx8640 systems. I am assuming that both rx8640 systems have been properly maintained and that the cmcluster directories are the same on both systems. If not, reconcile the differences before copying to the blade.

 

Edit the /etc/rc.config.d/netconf file to setup the network cards.

 

Install the latest version of SgeSAP to match the blade's patch level. You'll need to configure SgeSAP for the current environment.

 

http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=4&cad=rja&uact=8&ved=0CDgQFjAD&url=http%3A%2F%2Fh20564.www2.hp.com%2Fportal%2Fsite%2Fhpsc%2Ftemplate.PAGE%2Faction.process%2Fpublic%2Fkb%2FdocDisplay%2F%3Fjavax.portlet.action%3Dtrue%26spf_p.tpst%3DkbDocDisplay%26javax.portlet.begCacheTok%3Dcom.vignette.cachetoken%26javax.portlet.endCacheTok%3Dcom.vignette.cachetoken%26docId%3Demr_na-c02...

 

Run cmcheckconf on both the cluster config file (typically called ...ascii) and the package control files. You'll need to correct any errors before continuing. Be sure to handle setting up the shared volume group(s) and lock disk.

 

When cmcheckconf runs error free,make copies of the cluster and package scripts on the primary node, then copy the new scripts from the blade to the primary node.

Now stop the package(s) and the cluster.

Then run cmapplyconf to create the new cluster environment.

 

If all is error free, then start the cluster with cmruncl and check the status with cmviewcl.

 

I expect a number of failures will occur along the way since you are not following the recommended steps (upgrade SG, patching, etc). This is a senior level task with a lot of risks and downtime. If you are not familiar with Service Guard, I wouild not start the task without the guidance of a good consultant.

 



Bill Hassell, sysadmin
Mehul5
Frequent Advisor

Re: Cluster Migration

Hi Bill,

 

Thanks for the information.

 

Yes we  are running on the rx8640 DRD disk.

 

Also I want to inform you that we have already successfully completed the migration activity with almost the same procedure and also this migration was carried out by senior HP members at my site only but associated with different project.

 

Thank you very much for the valuable information and help.

 

Regards,

Mehul