Video Screencast Help

Windows 2003 cluster update (how to test a clustered backup setup)

Created: 28 Feb 2013 • Updated: 14 Mar 2013 | 23 comments
This issue has been solved. See solution.

Hello,

 

I ran the backups for the cluster setup (backups ran however not sure if it works)

Policy 1 >>>pagicianqa_clstr_physical    Backup seems to have  completed successfully

In the client selection I put the following full physical node names the backup selection for these are c:\ and system state.

pagicianq1.my.company.com  >>>completed successfully

pagicianq2.my.company.com >>>completed successfully

I also have  a third  client >>> pagicianqa_cluster  (not sure why I put that) I think that’s what I put as the client name when I was creating the policy. This is getting a netbackup code of (48) client hostname could not be found, which is correct since there is no client with that name so I guess I can delete that client name.

Feel free to ask questions, as I am not sure if this is correct or not.

Policy 2  clstr_pagicianq1 (which backups up the shared drives) >>>  H:\  Q:\   

UPDATE >>I just found out that the C:\ and D:\ drives are the physical drives so I  removed them from backup  policy  2 (the one that backups up the shared rescources). I also removed the I: drive the T: drive, E: drive, F: drive and the G:drive (per admin instructions)

So the backup selections for the shared/resources are the H:\  Q:\

I will add the D:\ to the policy 1>>pagicianqa_clstr_physical and remove it from policy 2>> clstr_pagicianq1.  Is this going to be an issue now that I didn’t originally have this in place?

Fyi –  both these policies/backups started at midnight  could that be an issue?  Also  how do we test this cluster setup to see if it works/worked!

Thanks in advance!

 

Discussion Filed Under:

Comments 23 CommentsJump to latest comment

Nagalla's picture

what is the client name that you are using for the policy clstr_pagicianq1?

does this shared resources are active -active or its active -passive?

means, both Nodes can access the H and Q drives same time or only one node can access it?

 

see the below tech note to configure the cluster Drives

http://www.symantec.com/business/support/index?pag...

nbustarter380's picture

Thanks Nagalla,

The name of the client for policy clstr_pagicianq1 is

Pagicianq1.my.company.com >>> this is the name of the the physical node that currently owns the shared resources/drives

Backup client shared drives H:\ and  Q:\

Should I  add the name of the 2nd node of the cluster >>>pagicianq2.my.company.com?

Active - Passive (per resource)

Tho I was informed to keep in mind  that Q and H can exist on different servers at the same time.

Thanks…I have that tech note

One of the parts I think I did incorrectly is below 

Configure the Schedules so that the Policies do not run at the same time as either the Virtual Server backups or the other cluster node(s) backup

I have my 2 policies running at the same time (they both completed last night)

Hope I don’t have re-do this whole thing!  And it's just a matter of adjusting things.  Keep in mind I have never done a cluster setup before.

Thanks again!

Nagalla's picture

hi,

you are missing one more thing form the T/N

for the cluster Drives, you would need to use the Cluster name as a client name.so that you backup always points the active node of the cluster.

 

  1. b) To backup the Virtual Servers:
  2.  
  3. Create an individual policy for each Virtual Server in the cluster
  4. Specify the Virtual Server hostname in the 'Client Name' field on the 'Clients' Tab.
  5. Specify the Shared Disk associated with the Virtual Server
  6. Select the Storage Unit or Storage Unit Group of the preferred NetBackup Media Server.
nbustarter380's picture

Thanks Nagalla,

So in my case the cluster name (for policy clstr_pagicianq1) would be >>>pagicianq1.my.company.com which is there already.

 

Or are you saying the cluster name should  also be pagicianq1.my.company.com  the same as the client name pagicianq1.my.company.com.?

Also what about the other node in this cluster>>> pagicianq2.my.company.com? should that be added also? The reason I asked is if for example

The H: drive is moved to pagaicianq2.my.company.com will know to it up on that node?

 

Best Regards!

 

 

 

Nagalla's picture

its pagicianq1.my.company.com one of the node name in your two node cluster, its not a culster name.

cluster name  ?

node 1:-pagicianq1.my.company.com

node 2:- pagicianq2.my.company.com

cluster name would be the one which always points the active node of the cluster.

so it the H drive(cluster resource) is in pagicianq1., cluster name  takes you  to pagicianq1.

if  H drive(cluster resouce) is in pagicianq2 cluster name take you to pagicianq2

and this clsuter name also should have assinged with IP and binded with cluster resource group.

please check with your cluster admin if you are not sure about this.. 

 

 

 

 

nbustarter380's picture

Nagalla Thanks!

 

Here is my setup,

 

Policy 1

pagicianq1_clstr_physical

under client name tab I have the following  these are the 2 nodes in the cluster

pagicianq1.my.company.com

pagicianq2.my.company.com

under backup selections I have the following

c:\

system state

d:\

 

Policy 2

Clstr_pagcianq1<<< I chose this as the cluster name

Client name tab >>>pagicianq1. My.company.com

Backup Selections

H:\

Q:\

So I think what your saying is, I should chose a different  cluster name  and IP address?

What I am also confused at to where is this new IP address is coming from?

 

so far I am using 2 IP addresses

pagicianq1.my.company.com >>>123.12.12.12

pagicianq2.my.company.com>>>123.12.12.12

 

Best regards

 

 

 

Nagalla's picture

correct,

you need to ask the cluster admin , for the cluster IP address and the Name, 

this cluster IP always points the active node of the cluster.

and you need to provide the cluster name in the backup policy  clients tab. so when backup trigger it always goes and communicate with the server on which the H and Q drives are active.

so  it does not matter which node having the drives H and Q,  always have the successfull backup irrespective which node it is.

 

nbustarter380's picture

Thanks Nagalla, That information helps a lot I'm understanding things better now.

I will update you later when I make the additions/changes.

One more thing the backup schedules for the 2 policies they should be different correct?

if one policy starts at midnight I should make the other policy start earlier or later correct?

Best Regards!

 

 

 

nbustarter380's picture

Hi Nagalla, Thanks for your responses

Here is the windows client cluster setup that I currently have, If anyone has any questions or see something that may be wrong let me know.

Policy 1  (this policy backs up the physical drives and system state)

Pagicianqa_clstr_physical

Schedules

Full and Incrementals > start at 2am

Clients tab

Pagicianq1.my.company.com 

Pagicianq2.my.company.com

Backup Selections

C:\

System state:\

D:\

 

Policy 2 

Clstr_pagicianq1 (backs up the shared resources which may move between the two servers)

Clients tab

Logclusterq.my.company.com

Loggroupq.my.company.com (quorum group)  q:\

Pagicianq1.my.company.com  (database group (H: drive)

Schedules start at midnight.

Backup selections tab

H:\

Q:\

Also, Since I have been updating this set (several times) will that cause a problem? If this current setup is finally the way it should be what about the backups that were taken for the past few days when the setup was wrong will that be an issue or no?

How do we test the cluster setup?

Thanks

Nagalla's picture

 

Logclusterq.my.company.com

Loggroupq.my.company.com

why two names? are both are cluster? if yes, are they both have the common nodes?

both have the common nodes pagicianq1.my.company.com, and pagicianq2.my.company.com, you does not needs to use 2 cluster names in the policy using once is more than enoubh.

if both the names having the different nodes, other then pagicianq1.my.company.com, and pagicianq2.my.company.com, then you would need to create another policy for that, do not mix with it.

why still Pagicianq1.my.company.com node name in the policy, you can remove it as you specified the cluster name. you does not need to keep Pagicianq1.my.company.com  in clients tab, if you keep this it will try to keep the duplicate backup.

 

the backup that you made sofar is still valid, untill they reach their expire date.. you can make use of them when restore require, but need to use node name to search the backup images.

nbustarter380's picture

Nagalla thanks, I changed/deleted the names for policy 2 is the new setup

 

Note: policy 1 is the same I believe that is ok?

 

Policy 1

Pagicianqa_clstr_physical (This policy backs up the physical drives)

Schedules

Full and Incrementals > start at 2am

Clients tab

Pagicianq1.my.company.com

Pagicianq2.my.company.com

Backup Selections

C:\

System state:\

D:\

 

 

Policy 2

Clstr_pagicianq1 (backs up the shared resources which may move between the two servers)

Clients tab

Pagclusterq.my.company.com (quorum group) q:\

Paggrouph.my.company.com (database group (H: drive)

Schedules start at midnight.

Backup selections tab

H:\

Q:\

 

Nagalla's picture

why two names in clients tab still

Pagclusterq.my.company.com

Paggrouph.my.company.com

does this both for cluster names one for Q drive and another for H drive having the same nodes Pagicianq1.my.company.com  and Pagicianq2.my.company.com ?

if yes. you would need to create one more policy.

see below.

Policy 1

Pagicianqa_clstr_physical (This policy backs up the physical drives)

Schedules

Full and Incrementals > start at 2am

Clients tab

Pagicianq1.my.company.com

Pagicianq2.my.company.com

Backup Selections

C:\

System state:\

D:\

 

 

Policy 2

Clstr_pagicianq1 (backs up the shared resources which may move between the two servers)

Clients tab

Pagclusterq.my.company.com (quorum group) q:\

Backup selections tab

Q:\

 

Policy 3

<policyname ) to backup up the shared Drive H for the cluster Paggrouph.my.company.com

Paggrouph.my.company.com (database group (H: drive)

 

Schedules start at midnight.

Backup selections tab

H:\

 
nbustarter380's picture

Nagalla thanks,

Does this both for cluster names one for Q drive and another for H drive having the same nodes Pagicianq1.my.company.com  and Pagicianq2.my.company.com ?

yes, that’s correct those are the shared resources that sometimes can move between Pagicianq1.my.company.com  and Pagicianq2.my.company.com

 

I created a third policy as you suggested (see below)

Policy 2

Clstr_pagicianq1 (backs up the shared resources which may move between the two servers)

Clients tab

Pagclusterq.my.company.com (quorum group) q:\

Backup selections tab

Q:\

 

Policy 3

>>Clstr_pagicianqh) this policy backups up the shared Drive H for the cluster Paggrouph.my.company.com

Paggrouph.my.company.com (database group (H: drive)

 Schedules start at midnight.

Backup selections tab

H:\

Let me know if this is correct? If so How do we test this cluster setup?

Best regards!

 

Nagalla's picture

looks fine to me..

go ahead and run the backup,and if you see the successfull backup, that indicates you did the successfulll configuraitons.

nbustarter380's picture

Thanks Nagalla for your help!

 

Question if the backup is successfull does that mean the configuration is correct as far as the cluster is concerned? meaning if the shared/resources moved to the pagicianq2? Is there some type of  failover that is done to test that?

Best Regards!

Nagalla's picture

yes,

first test the backup with current state and make sure that is successfull then 

you can ask your cluster admin to do the failover of the cluster resources, and test the backup again and see how the backup works... 

even aftar the failover your policy 2 and 3 got successfull to take the backup..that confirms that you hvae configured the cluster policy fine.

nbustarter380's picture

Nagalla Thanks,

The 3 policies completed successfully (policy 3 was a full since it was it's first time run). I will see if they run successful again and if so then we will test the failover.

Best Regards!

 

nbustarter380's picture

Nagalla , Thanks

 

The admin moved the files from H:\Archive to a H:\TempArchives.

Then he ask me restore the H:\Archives folder

I Files restored successfully

Then he moved the cluster group to >>Pagicianq1.my.company.com and I did the same restore again

The files were restored successfully.

Is there any other restore we should try? I just want to make sure everything is correct before we do the production windows cluster.

Thanks for your help!

Nagalla's picture

Nbustater,

you did the restore test which is successfull..

you are good to move them to Production...  :-)

SOLUTION
nbustarter380's picture

Great! Thanks Nagalla for all your help I really appreciate it.

One last thing if they wanted to add 2 nodes to the cluster in the future, I would just add them to the 3 policies following the same procedure correct?

Best Regards!

 

Nagalla's picture

are you saying about the another cluster or existing cluster?

if that is the new cluster , yes folllow the same process.

if that is for the existing cluster that you already configured.. then you does not need to add the node names in to the policy having the cluster drives... add them in normal policy  to take the backup of local Drives if require.

nbustarter380's picture

Nagalla Thanks,

I was asking about a new cluster however, I believe that 2 servers may be added to the existing cluster in the future. So thanks for the information on that configuration as well.

Best Regards