Ah clusters!  Backing them up can be a pain!

Well, not really if you do the following:

Set the cluster physical servers (OS level) to be backed up to the physical addresses and backup only the operating system drive

Set the VIRTUAL clients to backup to the virtual IPS and backup only application data drive(s).

This will mean you have at least  3 clients:  one for each physical server and one (or more) for the virtual

By doing this, the backups will better recognize the files that have been backed up.

Also, what is the retention for the backups?  That may cause an issue.

Finally, remeber that if the group has a schedule assigned to it, it wil override the clients schedule.  Be sure you do not have the group scheduled - or at least it is the same schedule the client(s) should have.

Mark

HI Mark

This forum is quite related to me !!

Kindly can u give us some tips to configure the cluster client backup  beacuse in our enviorement physical nodes backup is happening fine but cluster backup is failing.

Do we need to use the backup command i.e(save -c -a ?

Thanks

Hi Mark,

Thanks for your valuable reply.

We are having Networker server 7.6.2 and OS version: RHEL 5.6.

Retention period is 3 days. I have checked the Group Propertie's schedule in Advanced tab. Both the schedules are same. Every hour we are taking log file backup.  Now, i am about to try your suggestions. I will let you know the output.:)

-Sabarna

Hi Mark,

As per our configuration in the Networker, we are having multiple clients in that group where i tried your suggestions but i selected one local file system (/tmp) for physical ip and application data for Virtual IP but still full bkp is happening. plz suggest.

Sabarna

Hello Sabarnana

It would be worth testing the virtual node backup using a small saveset so you can allow the full backup to complete and then

To confirm which actual client is attached to the test backup , please check using the name of the physical nodes and virtual nodes

example:

# mminfo -q client=physicalnode1  -r name,level,savetime

# mminfo -q client=physicalnode2  -r name,level,savetime

# mminfo -q client=virtualnode  -r name,level,savetime

Also in the indexes directory on your networker server.

before the backup do 

ls -l /nsr/index/physicalnode1/db6

ls -l /nsr/index/physicalnode2/db6

ls -l /nsr/index/virtualnode /db6

Check these again after the test backup of the virtual node and confirm which directory has been updated most recently.

Regards,

Conor.

Yes i got the solution.

1>On both Physical servers, create pathownerignore file under save binary directory. (ex:/usr/sbin)

2> On Virtual client properties from networker, type "save -c " in backup command.

But i need to check whether it will effect clustering or not. I will let you know the result.

That's not solution.  pathownerignore says you will backup cluster owned files by doing backup of physical box. 

What you should do is keep -c as you did for virt, remove pathownerignore (I believe you need to restart nsrexecd to take affect) and manually create lcmap file (manually because there is official support of RedHat cluster so there is no support for it in networker.cluster - so unless you run VCS or EAS you will need to do that manually).  Then, saveset all against virtual client will work too (but I recommended to keep -c nevertheless for virtual client).